[Last 50 Posts]
This thread is for the discussion of cybercommunism, the planning of the socialist economy by computerized means, including discussions of related topics and creators. Drama belongs in /isg/
ReadingTowards a New Socialism by Paul Cockshott and Allin Cottrell:
http://ricardo.ecn.wfu.edu/~cottrell/socialism_book/Brain of the Firm by Stafford Beer
Cybernetic Revolutionaries by Eden Medina
Cybernetics: Or the Control and Communication in the Animal and the Machine and The Human Use of Human Beings (1st edition) by Norbert Wiener
Economic cybernetics by Nikolay Veduta
People's Republic of Walmart by Leigh Phillips and Michal Rozworski
Red Plenty by Francis Spufford
Economics in kind, Total socialisation and A system of socialisation by Otto Neurath (Incommensurability, Ecology, and Planning: Neurath in the Socialist Calculation Debate by Thomas Uebel provides a summary)
Active writers/creatorsSorted by last name
>Paul Cockshotthttps://www.patreon.com/williamCockshott/https://www.youtube.com/channel/UCVBfIU1_zO-P_R9keEGdDHQ (
https://invidious.snopyta.org/channel/UCVBfIU1_zO-P_R9keEGdDHQ)
https://paulcockshott.wordpress.com/http://paulcockshott.co.uk/https://twitter.com/PaulCockshott (
https://nitter.pussthecat.org/PaulCockshott)
>Cibcom (Spanish)https://cibcom.org/https://twitter.com/cibcomorg (
https://nitter.pussthecat.org/cibcomorg)
https://www.youtube.com/channel/UCav9ad3TMuhiWV6yP5t2IpA (
https://invidious.snopyta.org/channel/UCav9ad3TMuhiWV6yP5t2IpA)
>Tomas Härdinhttps://www.haerdin.se/tag/cybernetics.htmlhttps://www.youtube.com/channel/UC5fDgA_eHleDiTLC5qb5g8w (
https://invidious.snopyta.org/channel/UC5fDgA_eHleDiTLC5qb5g8w)
>Elena Vedutahttp://www.strategplan.com/en/about/veduta.phpVarious videos on YouTube but no channel of her own
>Dave Zachariahhttps://www.it.uu.se/katalog/davza513One video on Paul Cockshott's channel
Podcasts>General Intellect UnitPodcast of the Cybernetic Marxists
http://generalintellectunit.net/Previous threads in chronological orderhttps://archive.is/uNCEYhttps://web.archive.org/web/20201218152831/https://bunkerchan.xyz/leftypol/res/997358.htmlhttps://archive.ph/uyggphttps://archive.is/xBFYYhttps://archive.ph/Afx5ahttps://archive.is/kAPvR>>1650427
>>1852043>This thread is for the discussion of cybercommunism, the planning of the socialist economy by computerized means, including discussions of related topics and creators. Drama belongs in /isg/Should have included the LTV in your first paragraph.
>>1852050as much as I despise rightoids and wish to see them suffer, bullets are cheap
>>1852055someone earlier in the last thread said the consensus was to get rid of Victor Mangarino or w/e and basically the thread should be about planning, honestly the LTV and marxist/marxian econ in general deserves its own thread. If you're going to include mangarino you may as well include Anwar Shaikh and others like Ochoa.
>>1852065>cyber planned econ w/o LTVGood luck, I guess.
>>1852072just because its not in this specific thread doesn't mean people wont read about it lol. This is a thread, not a university degree curriculum
>>1852065>>1852072actually a Shaikh/Magariño thread sounds like a good idea. the spat between them and Cockshott is interesting and worthy of a thread of its own
Cybercom is becoming highly relevant no thanks to dogmatic retards but because of technological development.
>>1852094this raises a question: are there anyone besides cybercommunists seriously advocating planning?
>>1852043So, do you prefer CPLEX or Gurobi? I think it's pretty exciting that optimisation software has gotten so good that there is no longer need for specialized software for economic planning.
>>1852129who doesn't? any communist since marx advocates planning. the only leftists who don't are uh.. proudhonists, 'market socialists' and modern socdems (not leftists)
>>1852129The cybernetic part of planning is much less important than the democratic part in my opinion. You can create a functional system without computers but not with unaccountable self serving bureaucrats.
>>1852217>than the democratic part in my opinionIn decentralized planning?
>>1852219I guess I don't really understand how decentralised planning would work.
>>1852217the democratic part is kind of central to the cybernetic part though. for the system to work well you want transparency. you also need to automate the coordination aspect of it. part of the reason the Soviet system ended up the way it did was the computational-technical limitations of the 1920's when the system was set up. these days it would be obviously silly to set up a system where people would travel with briefcases of statistics and adjusted plans (as Red Plenty suggests it was done)
in the minds of liberals, Gosplan is imagined to be some kind of nefarious organization, with those at the top supposedly being particularly bad. but if we read the Kossov interviews we learn that the people at the top truly wanted the system to work. Baibakov was nothing if not loyal, and as an oil man, wanted the petroleum industry to work well. the same seems to apply for all higher-ups
>>1852220most people who propagate "decentralized" planning never seem able to separate the computational-technical aspect of it from the political. some even seem bent on recreating the problems of Gosplan by putting a bunch of middlemen in the system. if the computational part of it is to be decentralized then this seems to imply there should be some kind of corporate secrecy
>>1852386>never seem able to separate the computational-technical aspect of it from the politicalThat's a hard problem.
"Conceived most superficially, distribution appears as the distribution of products, and thus further removed from production and quasi-independent of it. But before distribution becomes the distribution of products, it is (1) distribution of the instru- ments of production, and (2) (which is another determination of the same relation) distribution of the members of society among the various types of production (the subsuming of individuals under definite relations of production). The distribution of products is obviously merely a result of this distribution, which is comprised in the production process itself and determines the structure of production. To examine production separately from this distribution, which is included in it, is obviously, idle abstraction; whereas conversely the distribution of products is automatically given by that distribution, which is initially a moment of production. Ricardo, whose object was the understanding of modern production in its specific social structure, and who is the economist of production par excellence, for this very reason declares distribution, not production, the proper subject of modern [political] economy. This is added proof of the absurdity of those economists who treat production as an eternal truth, and confine history to the domain of distribution." Marx,
Economic Manuscripts, Introduction, section 2b, "Distribution and production"
"The structure of any system designed by an organization is isomorphic to the structure of the organization." -Yourdon and Constantine,
Structured Design>if the computational part of it is to be decentralized then this seems to imply there should be some kind of corporate secrecyTrue, and bureaucracy and secrecy are I keep wondering just what it is people mean by decentralization, with so many contradictory prospects. One benign reading is that all planning data is public and anyone who cares to can replicate the calculation on personal hardware, propose their own what-ifs against the model, and so on. That's a minimum case to claim worker supervision of the production system. Another is to capture surplus compute cycles before they perish with a GOSPLAN@Home sort of thing, which could be a harmless gesture of civic participation and pride, albeit technologically unnecessary and possibly economically counter-productive. Finally there are concerns about partition tolerance, when lines of (informational and material) communication fail, especially under hostile conditions (nested control loops or Kalman filters might be worth exploring, with attention to boundaries of firms). But other rationales for decentralization seem to further the separation of labor from its conditions and risk pressing production into the defense of partial social identities. All that said, it may be prudent not to ship 50kg of hydrazine to just any residential apartment
>>1852087>the spat between them and Cockshott is interesting and worthy of a thread of its ownnot really its a pretty minor disagreement, a thread for that is an overkill, are you going to have a thread for every minor arcane issue in marxist economics, especially when most people here don't even have a grasp of the basics, or even have read volume 1? Better to just have a master thread of marxian econ
>>1852129>are there anyone besides cybercommunists seriously advocating planning?real life capitalist economies already have planning. What capitalism is is skewed by large western free market economies like the USA, which btw still has a ton of planning due to a large military/defense sector and government contracts, essentially practicing military keynesianism.
Most capitalism, especially in smaller countries where free market doesn't work, has some level of planning to even make shit work at all. capitalism with zero government planning or intervention hasnt worked since the late 19th century. This is where american right winged libertarians are actually correct except that they see it as some sort of conspiracy rather than a rational adjustment of the system to actually stabilize it. So liberals and dirigisme advocates are actually more realistic capitalists because they saved the irrationalities of the system from causing it to self destruct. If Keynesianism (and similar theories) hadn't been invented capitalism would have already died by the 1930s/40s.
>>1852533I think you are underestimating the degree of planning that happens within a large enterprise or a conglomerate, investment banks, the capital class of professional managers and corporate board members directing a dozen firms each, and the private intelligentsia.
>>1852556yes because in the US you have large corporations like walmart/amazon which are big enough to eventually replicate the sort of wide scale planning done by governments in smaller countries in which case its in line with the neoliberal philosophy of privatizing every thing including wide scale planning of society and a whole countrys political economy especially with something like amazon which is becoming the de facto "everything store" for all of north america
>>1852143thats not really the same as planning software, the fact that numerical linear algebra libraries exist isn't new, there will still need to be a whole lot of shit on top of that, private cloud infrastructure, database, UI and analytics/visualizations etc for people to input stuff into.
>>1852514>One benign reading is that all planning data is public and anyone who cares to can replicate the calculation on personal hardware, propose their own what-ifs against the model, and so on. That's a minimum case to claim worker supervision of the production systemyes this would be desirable, and the proposals for "decentralized" planning seem designed to make it impossible. or maybe I'm misreading them. a lot of it just seems to be blathering on endlessly about nothing, or criticizing some strawman version of what Soviet planning was like, as opposed to the actually existing system of planning (of which no coherent model exists, nor arguably could be built)
>Another is to capture surplus compute cycles before they perish with a GOSPLAN@Home sort of thing, which could be a harmless gesture of civic participation and pride, albeit technologically unnecessary and possibly economically counter-productivethis is an intriguing idea. I suspect the amount of data might be difficult to mirror on consumer grade hardware. it might be better to think in terms of there being a cloud which users can interact with, and to not be so concerned with where the physical machines are located
>Finally there are concerns about partition tolerance, when lines of (informational and material) communication fail, especially under hostile conditionsthere is robust theory in the database world on how to deal with problems like these
One problem I havent seen discussed here is deplatforming and proprietary software. Assuming socialism isnt global and goes in one country at a time, that country will likely be heavily sanctioned and their access to technology cut off. So any public cloud ex: AWS would be unavailable to them.
So in addition to making their own software they'd have to make their own IT infrastructure, assuming they dont start getting sanctions on servers, routers, etc. as well.
>>1852647>So in addition to making their own softwaremuch of the West runs on free software
>their own IT infrastructurethis goes without saying tbh
>>1852217One common proposal to democratic planning is to have a simulated supply and demand modifications, e.g. if a product runs out during planning period n then in period n+1 X% more of the product is made.
>>1852585>designed to make it impossibleFrom your account, it does sound like they tend to posit a human in the loop for bureaucracy's sake. It's an interesting way to institutionalize a mass line for those who believe such an institution is important.
>not be so concerned with where the physical machines are locatedAs for the front office, sure. But there is also a back office that must not fail to keep those square meals moving. One of the arguments for decentralization is that the lack of any justification to keep a central archive of the production conditions and worker records of every single piece ever produced. I envision full worker traceability of a particular copy of a simple consumer commodity like the grabby-grabby consuming over a hundred kilobytes down to its components, easily. That's a hell of a lot of data to fly from one side of the globe to the other every time a mold opens up or a tonne of white salt comes out of the reactor.
On the other hand, there is a
social argument for keeping a full history of the production of individual pieces. The market erases the prints of all the hands that produced a commodity. Cheap af storage and laser marking are broadly applicable ways to negate that negation for many products. The ability for someone to point to an object and know their own part in making it makes the production process individually meaningful.
>there is robust theory in the database world on how to deal with problems like theseIIRC the big one says that when communications are partitioned because of disaster or what have you, and a region's economy isn't partition-tolerant with full warehouses and local energy and water supplies, our preference for consistency or availability, as embodied in the design of the economic automation system, will be tested in its material implications. But events don't necessarily fall along lines we've set for them, so whatever you do, don't sacrifice availability.
There's been a lot of work privately put into commodity production logistics in China that could be nationalized when Xi pushes the communism button. I think Shein is an interesting case study. They track real demand from sales and automatically reorder in demand items from manufacturers. However it is limited as an example because clothes are cheap, short turnaround products. This system is only democratic in the "vote with your wallet" sense.
>>1853679>64:27 aspect ratio>(4:3)^2 = 16:9>(4:3)^3 = 64:27wdhmbt
>>1852217The actual democratic mechanisms haven't been properly researched. I've seen Cockshott talk about a system his students developed called HandiVote or something, that uses SMS for voting, but that has a lot of problems. It would be extremely easy to match a voter with their vote and so on.
I haven't seen any other attempts at imagining how such a system of direct democracy would work. It lacks a ton of detail.
>>1853778it's ez if you ditch vote secrecy. to not stand by one's votes is libshit
>>1853778>It would be extremely easy to match a voter with their vote and so on.thats the point, anon
>>1853521>It's an interesting way to institutionalize a mass linethe Soviet experience indicates the opposite. you can maybe do better by replacing these people regularly. I'd argue it's better to remove them completely. whenever possible, give your job away to a machine
>One of the arguments for decentralization is that the lack of any justification to keep a central archive of the production conditions and worker records of every single piece ever producedyeah but that has fuckall to do with centralization vs decentralization. it's possible to have a centralized database that is regularly pruned, and to have a decentralized set of databases that are never pruned. these people don't know what they're talking about
>That's a hell of a lot of data to fly from one side of the globe to the other every time a mold opens up or a tonne of white salt comes out of the reactorsure, but we can provide condensed views of the fully disaggregated data. data that may otherwise, in the worst case, be sent via airplanes loaded with tapes (as in VLBI astronomy)
pulling some random numbers out of my ass, the day-to-day data is probably more than 1 KiB/person/day but less than 1 MiB/person/day. with a working population of 5 billion that's 5 TiB/day. well within the capabilities of a regional data center, which in turn can churn out relevant statistics to anyone who asks. we could also implement internet-wide broadcast for this stuff, were the internet in proletarian hands
>The ability for someone to point to an object and know their own part in making it makes the production process individually meaningfulthat's a compelling counter to the usual whining about automatic planning somehow causing alienation. there seems to be an endless parade of "socialists" who aren't so interested in actually socializing production, with all that entails
>IIRC the big one says that when communications are partitioned because of disaster or what have you, and a region's economy isn't partition-tolerant with full warehouses and local energy and water supplies, our preference for consistency or availability, as embodied in the design of the economic automation system, will be tested in its material implications. But events don't necessarily fall along lines we've set for them, so whatever you do, don't sacrifice availabilityI think you can achieve both by using something like git (though git's porcelain sucks for this application)
>>1853806I think it's more to prevent people being intimidated or bribed for their votes but I get what you're saying. I still think voter secrecy for citizens is preferable but if it's impossible to do direct democracy without revealing people's votes then fine.
>>1853806>>1853812I don't agree at all. Voter secrecy is fundamental in a democratic society. Not having it will lead to coercion and a manipulation of the public's opinion and this will lead to the restoration of capitalism.
>>1853818>you can maybe do better by replacing these people regularlyI only saw an oppportunity to leave a point for people more theoretically advanced than myself to inject any human dependencies their theory requires.
>whenever possible, give your job away to a machineBased, 100%
>these people don't know what they're talking aboutT-that was me… and yes. I was concerned as much about network bandwidth, CPU, and that writing twice and reading never isn't entirely cost-free especially on solid-state media, as about storage per se.
>5TiB/dayIn that case, not such a big deal. Indexes might push that up by an order of magnitude but some govt agencies pay a couple million bucks a year for pens and paper. However, there are other personal imputations outside the factory proper, mostly to do with consumption metering or traffic control, probably mostly inherited from the capitalist order and (I expect) used to inform production or progressively scrapped.
>I think you can achieve both by using something like git (though git's porcelain sucks for this application)MapReduce seems like a good fit for a ledger, and some NoSQL databases distribute nicely.
>>1853935That sortition is the more communistic mode of selection has been deliberately forgotten in the Modern era.
>>1853951>T-that was mes-sorry
the point there is you need to be very precise what you mean. what exactly is to be "decentralized"? it could be argued that full centralization is precisely what is necessary to achieve what those advocating for "decentralized" planning hope to achieve. this misuse of words is of course highly irritating. we need better language
>However, there are other personal imputations outside the factory proper, mostly to do with consumption metering or traffic control, probably mostly inherited from the capitalist order and (I expect) used to inform production or progressively scrappedI'm certainly not keen on keeping the surveillance capitalist BS in place. we can have distribution centers ("stores") collect only daily aggregate statistics rather than personal statistics
>MapReduce seems like a good fit for a ledger, and some NoSQL databases distribute nicely.I think MapReduce would rather be for computing statistics, no? as for NoSQL, a lot of this data is tabular. notably, ledgers
I've made this to show how workers could arrive to a plan. When reading Isaac Rubin, he talks about relations, like indirect or market and direct, so…
Then I recall reading Dunayevskaya, something about workers and non-workers plan.
All in all, it ends with workers control, they have it or not, as I see it.
In the news everyday there is something about anthropogenic climate change. It is written as it is people doing. But I have no control like on the pic. It is not people, it is bourgeois. Planning like this can't be done with bourgeois controlling things.
Idk, may be address people directly, how they could change things if they had structures for that.
>>1856049> Planning like this can't be done with bourgeois controlling things.not only you can't but in fact free market liberalism completely relies on the impression that you can't plan for stuff on a macroeconomic level, it's very silly.
>>1856278it's going to be visual basic since it's compatible with excel macros, comrade.
>>1856278More importantly, how are existing ERP systems inappropriate to the task? Material reasons only please.
>>1856384I don't think anyone's done any evaluation of existing ERP systems
>>1856278>what programming language would an ERP system be written in?What a retarded discussion tbh.
Whenever you start writing, just pick something applicable at the time
>Relatively modern with existing long term support, so no new meme languages>Easily maintainable>Safety features>Not too hardware specific>Build system is FOSS and checkable.Probably some flavour of C then.
>>1856954>Probably some flavour of C then.Kek
>>1856954>C>Relatively modern>Safety FeaturesI lold. Truly deranged LARP if you weren't joking. Business software is written in the flavor of the month because its picked by managers based on the labor supply. Maybe in the 80s it would have been C, but now it's probably Java or C#.
>>1856998I was making a joke since the choices when having to pick something with long term proven support as well as open source compilers (which can be forked and verified in a seperate version) to prevent foreign sabotage doesn't leave a lot of languages to choose from.
You cant exactly pick Java because then the CIA can just insert backdoors into the runtime or compiler via oracle, same for C#.
>>1857027(and if we're going to write our systems in javascript I will start protracted peoples war against Gosplan)
>>1857031What about the five year plan to transition to Haskell?
>>1857027>long term proven supportYour jokes only get less funny. Stop posting.
>>1856998>I lold. Truly deranged LARP if you weren't joking. Business software is written in the flavor of the month because its picked by managers based on the labor supply. Maybe in the 80s it would have been C, but now it's probably Java or C#.C++ while using modern features is relatively safe ex: smart pointers instead of raw pointers. Its nearly as good as garbage collection and on par with languages like Swift.
>The C++ smart pointer idiom resembles object creation in languages such as C#: you create the object and then let the system take care of deleting it at the correct time. The difference is that no separate garbage collector runs in the background; memory is managed through the standard C++ scoping rules so that the runtime environment is faster and more efficient.People think C++ is C, or that C++ is unsafe because they were taught to use raw pointers in the DS&A class by a professor teaching from an ancient textbook. The fact is modern C++ is almost as safe as managed languages like Java/C# while being a hellof alot more performant which is why its heavily used by Unreal Engine, Embedded Software, etc.
>>1857049In all likelihood most planning software will probably be written in Python
or some similar "light/scripting" language due to its familiarity to non developers like data analysts, scientists, economists, etc. Anything performance intensive will just be a library written in C/C++/Rust/whatever and called by a Python Script (ex: PyBind11, ctypes).
Those claiming C is insecure are codelets who don't know about formal methods and C interpreters
>>1857141He said "some flavor of C" not C++. Even with RAII and Smart Pointers C++ isn't particularly memory safe, or more importantly ergonomic.
>>1857194>formally validating a C program>interpreting CCniles will do anything to avoid learning a language designed after 1980
>>1856998>The C++ smart pointer idiom resembles object creation in languages such as C#: you create the object and then let the system take care of deleting it at the correct time.Tracing GC and reference counting are absolutely not alike, only at a surface level if anything.
Many GC implementations do not allow destructors, meanwhile with reference counting you can have egregious memory leaks if the object graph ever contains a cycle.
>>1857141> People think C++ is C, or that C++ is unsafe because they were taught to use raw pointers in the DS&A class by a professor teaching from an ancient textbook. The fact is modern C++ is almost as safe as managed languages like Java/C#Absolutely not. C++ is still ripe with footguns thanks to the standard library, and while the situation has improved, you still have nonsense like null unique_ptr (which must be this way because of movable objects).
On top of it, as far as building systems go, C++ is straight up nightmareish. CMake (even "modern" CMake!) is a fucking mess, and including any dependency beyond a header file is a straight up nightmare.
I personally spent less time rewriting a whole project from C++ to Rust than what I've spent trying to include a CMake C++ library in my CMake C++ project
>>1857194> Those claiming C is insecure are codelets who don't know about formal methods and C interpretersShitty cope coming onky from retards who haven't wrote a single line of code in their lives or idiots with their heads shoved so deep up their arses to be beyond repair.
It has been demostrates innuberabke times that ¾ of bugs in software projects like chromium, windows, linux etc come from memory safety issues, which can be prevented by simply using a memory safe language (or a language that at least it's safe by default like Rust).
And how it could have been otherwise? Machines are always, systematically more rigorous than any human could ever hope to be, even nore so an overworked wagie who has been staring at code for 5+ hours.
The same pattern repeats in literally any other industry and field, quit it with this senile hogwash nonsense
>>1857207those footguns will always exist because any hardware interactions and interactions with other libraries will involve unsafe rust anyway
>>1857311 (Me)
Oh, and there is apparently work being done to reimplement P9 in Rust which is pretty cool I guess:
https://github.com/dancrossnyc/r9
>>1857311its unclear why you need a whole OS for distributed computing when you can just write a program that does the same. After all, all kinds of corporations are writing microservices and other distributed systems out the wazoo which are communicating using grpc or MOM/middleware or http with xml/json or even plain binary TCP.
>>1857311>cyberplanning would rely on multiple computers operating in tandemWhy would you reproduce hierarchy without a material reason?
>>1857318Yes. But the advantage P9 offers is that it unifies all the ways computers communicate with one another into a single unified filesystem instead of this hodgepodge of arcane communication protocols, which makes the process of integrating and adding new computers to the planning network almost trivial (at least theoretically, there is a lot of work that's needed to bring it up to speed with modern hardware)
>>1857319How would it do that, though?
>>1857311To what degree? For computation you wanna have a massive server farm with the best GPUs money can buy
>>1857339>How would it do that, though?Presumably these systems would correspond to some material relation, right? The federated republic model might not be the actual model followed by a communist revolution.
Pdf shows run times of planning code no later than May 2020. I don't see hardware listed, but vanilla numpy can't into GPU so these were probably done on a regular CPU of the era. The graphs suggest that a plan for a village of several thousand takes about 2×10^11 operations at most to compute. Modern GPUs perform over 10^13 operations per second. Compute isn't the limiting factor (but maybe IO and RAM could be).
>it unifies all the ways computers communicate with one another into a single unified filesystemOnly if they're running Plan 9. Remember that the planning system will not be born into a pristine world organized according to post-capitalist relations. Its operators will be dealing with a hodgepodge on day 0 anyway and every day thereafter as more capitals are annexed from a foreign country called "Ohio steel capitals #4" and added into the planning system as means in detail.
Finally, a lot of work has been done on distributed computing in the Internet (and IoT!) worlds while Plan 9 has been sleeping. A plethora of distributed databases, fleet orchestration, publish/subscribe all bring a more autonomous means closer to reality, platform-independently.
>>1857447tons of programming language have nullable pointers ex: Java Optionals are nullable so the wrapper thats supposed to tell you whether something is null or not can itself be null. again, I don't know Rust but the only other language that comes to mind that has true non-nullable/optional types is Swift. Nullable pointers are quite literally then norm for most languages and even a ton of languages which tacked on optionals after the fact have the same problem as Java. Rust fanbois need to stop talking as though Rust = all "other programming languages".
>both will kill each other and we will all use Go instead
I've tried a bit of Go, made a dynamic array with malloc in C-go that can either be heap allocated or a Go slice and it outperforms standard go slices by huge margins, especially when the size of the data you're gonna put in the dynamic array is unknown.
For me Go was designed by great engineers, but drunk.
>>1857449right, I dont really understand why it wouldn't just be better using a server farm with openstack on top
anyways the thread has been derailed so much. every political talk online will devolve to comparisons to hitler, and every dev discussion online will devolve to language fights.
Use whatever, just not Python or PHP.
>>1857490uh… at risk of derailing further, why not python?
>>1857490Nope, Python is scientific and we are scientific
i seriously hope everyone who still has fights over what brand of the same tool they like the most is under 25
>>1857507its all in good fun anon
>>1857536if thats the case then its fine, im just used to legit fights where people get seriously into it, even in my workplace
>>1857552well,
there is a legitimate debate to be had on the pros and cons of what tool to use to write new software. It's just one of many things that does actually have to be decided at some point, in addition to a whole bunch of other stuff.
The difference is in a work environment its not just about the pros and cons of the language itself but the "ecosystem" of libraries around it for your use case, how many people at the org/company already know the languages and/or how easy it is to hire devs for that language. And of course, getting paid support doesn't hurt either so risk averse management has someone to blame if something goes wrong.
All that put together explains why companies are still using python, C#, Java 8 from 2014, and MAYBE golang/scala if they're edgy. The fact is companies other than startups are never going to use some hipster/meme stack that users of hacker news love, because they just want something that works, is proven to scale, and allows them to get a herd of middling programmers to churn out software that fulfills basic business requirements without shooting their feet off. The fact is everyone loves to be a hipster but 99.9% of run of the mill business backend web software can be done using basic bitch languages easily understandable by anyone so there's no reason to do anything else other than niche uses.
>>1857559makes me think of many small companies ive worked for wasting money on services like AWS when their size and requirements absolutely are nowhere near the level of the big guys and i have to hold back on asking the higher ups why the fuck theyre spending money that could go to better things (like our fucking wages)
one of the upsides of this is languages like cobol still being used today by pretty much all banks and even paid handsomely because nobody wants to learn that language and develop with it. its easy money but i hate working for banks so i dont botehr (maybe i should…)
>>1857565most cloud migrations dont save money anyway thats largely just AWS propaganda that people who take their certifications are brainwashed into. Basically people are getting a certification in how to be an unpaid AWS salesman in their own company.
>>1856049Do you think this could replace Cockshott, Marketing of consumer
goods?
The equilibrium is achieved not by adapting to sold/labor spent -> 1,
but by users themselves changing jobs or working differently.
So for milk, the estimate function should return 1, or there will
not be enough milk for those who consume it. For mobile phones
estimate can be 5 years, a phone will be replaced once in 5 years.
There is no vouchers nor money. If a user wants more things he can
work more and see effects not just in aggregates. May be something
like this:
u1, u2, .. - users. Lets virtually remove u1 from a factory and see
how estimate changes, record this. etc, for u2, … This will be an
ordering set, like, who first gets a car.
>>1857204C is the lingua franca of the computer world, and will remain so for many decades to come. Cope and Ceethe
>>1857207>Shitty cope coming onky from retards who haven't wrote a single line of code in their livesstop projecting so hard
>using a memory safe languageC is perfectly memory safe. all you need to do is define the machine appropriately. you pay for that memory safety in decreased performance however, which is unacceptable in many applications. HPC and video come to mind. you can write a BLAS implementation in Java but it won't be fast
people who whine about C are like people who whine about XML. junior developers who have not learned enough discipline
>Python is scientific and we are scientific
Python is glue, CUDA is scientific
>>1858068>junior developers who have not learned enough disciplinemost 21st century developers never even touch C anon, other than embedded and OS devs
>>1858068Boomer theology is boring, kys for piety
>>1858087Glue is society.
>>1858130that's fine. personally I prefer Python, and handing off any heavy calculation to C libraries for which Python bindings often exist. one prominent example is numpy (and scipy)
as far as I'm aware there's no hardware support for bounds checking. as soon as one allows indirect addressing you will get the kind of bugs we see in many C programs. the way to prevent these are either to do what Java does, namely bounds checking on every access, or somehow guaranteeing out of bounds access can't happen via the type system. formal methods allow the latter, even in C
>>1858141C will remain the standard for decades to come. cry all you want. only C is performant while also having a stable ABI. C++ does not have a stable ABI and neither does Rust (last time I checked)
>>1858149>Python, and handing off any heavy calculation to C libraries for which Python bindings often exist. one prominent example is numpy (and scipy)This will likely be the solution.
>>1858149>C will remain the standard for decades to come. Doesn't matter. You should be programming to a high level and compiling that into something else anyway.
>>1858176you
should use whatever tool is suited for the task. sometimes that tool is C
>>1858068I whine about C because I know from experience that things can be better.
Strong typing, proper modules, and generics make things significantly easier at absolutely zero cost, being resolved at entirely at compile-time. If you're willing to sacrifice a bit of simplicity, you can also throw operator and function overloading on the list.
>>1858280>he doesn't use multiple tools at onceAutism or religion?
what about cython? compiled python to C for performance.
https://en.wikipedia.org/wiki/Cython
>>1858339That's a deployment option.
>>1858339there's also pypy
>>1857473i heard rust is bad for game dev
>>1857498Shit tier performance for computer.
Also, this might be useful
A programming language for massively parallel conputation
https://github.com/HigherOrderCO/Bend
>>1858339>>1858493Still garbage collected with no option for cache layout optimization, no parallelism, etc
>>1858724The program finishes in like six seconds. Stop creating superfluous labor time, capitalist
>>1858068> C is perfectly memory safe. all you need to do is define the machine appropriately.Absolute retarded nonsense, and I have already explained why. Kick that damn bucket you boomtard
> you pay for that memory safety in decreased performance however, which is unacceptable in many applications.This moron has never heard of Rust
>>1858726In 6 seconds on what hardware, with what and how many inputs?
>>1858728<turns up to aggressively whine about programming shitYou people are mentaly broken past help.
I'm glad I cut this nonsense out of my life and got a normie trade job.
>>1858730Here is timing information from no later than May 2020,
no GPU because numpy doesn't support GPUs, a five year plan for a district of 10k enterprises. Let's assume the fastest DDR4 at 25.6GB/s per stick, 4 sticks, memory-bound. Assuming there exists a PC with a 102.4GB/s memory controller, in six seconds about 0.6 terabytes of data could have moved, / 4 for float32_t = 150Gfloat, theoretical, at most. I'm sure these districts can be composed into a second, whole-world planning run. I don't think it's necessary to consider the scarcity of computing time as a pretext to play around in C
but if you can write CUDA nobody will turn down your contribution
>>1858728> you pay for that memory safety in decreased performance however, which is unacceptable in many applications.> This moron has never heard of RustRust has to copy a lot of stuff, needs to initialize everything, everything ends up being in an Arc<RwLock<Mutex<Box<dyn FearlessConcurency>>>> if you dont have 10+ years of experience wrestling the compiler with lifetimes…
Rust is one of my favourite languages, but what i write in Rust, for peak performance, is not idiomatic nor memory safe, and for that it's easier to use an unsafe by default optimized language like C++.
I don't see Rust stealing C++ or CUDA's spot for high performance computing, maybe it could steal a bit of Java's for mid-high performance stuff.
>>1858728> you pay for that memory safety in decreased performance however, which is unacceptable in many applications.> This moron has never heard of RustRust has to copy a lot of stuff, needs to initialize everything, everything ends up being in an Arc<RwLock<Mutex<Box<dyn FearlessConcurency>>>> if you dont have 10+ years of experience wrestling the compiler with lifetimes…
Rust is one of my favourite languages, but what i write in Rust, for peak performance, is not idiomatic nor memory safe, and for that it's easier to use an unsafe by default optimized language like C++.
I don't see Rust stealing C++ or CUDA's spot for high performance computing, maybe it could steal a bit of Java's for mid-high performance stuff.>>1858728
Oops i posted twice, there were delays
>>1858799>maybe it could steal a bit of Java's for mid-high performance stuff.i dont think fintech bros are gonna learn rust. Golang is a far more likely candidate
>>1858805>materialism is when you quack opinions in a confident, businesslike formNo more sperging until you answer
>>1858744 and tell us why exactly your need to do work that has been automated out of existence should take up any time on the agenda. Lurk until you have ported
https://github.com/ssamot/socialist_planning to C and know what you're talking about
>>1858721interesting. but that GPU example doesn't look so impressive, a mere 57x speedup from 16k cores is 0.35% efficient. probably useful for high-level ||ism doe
for HPC, C+MPI is likely to remain king
>>1858728>Absolute retarded nonsenseI suggest you actually familiarize yourself with C. nothing says you have to compile it down to unsafe code. it can just as well run as a REPL with bounds checking etc
>This moron has never heard of RustRust doesn't solve the issue of bounds checking costing time. as
>>1858800 points out you have to use unsafe code for that
>>1858988Bound checkimg isn't really an issue 90% of the time. Just use iterators
>>1858988just graduated with a fucking bachelor's in computer engineering and I don't know what any of this shit means. what a stupid fucking waste of time and money
>>1859079iterators help, yes. but even Java has been shown to have OOB issues. relatively recently it was shown that its quicksort implementation breaks when using arrays of size 2^31 or so
claims about security are nothing but feels if they're not backed up by formal proofs
>>1859084>just straight up admitting to being a junior dev
>>1859084You're not missing much. What we have here is a classic case of angel-pinhead theorizing and some sperg trying to make other people play his autistic games with him.
Hide, ignore, etc.
>>1859095And? Nobody said that memory-safe languages are 30.000% bug free, but by virtue of being memory safe, be it with static analysis or runtime checks, they straight uo eliminate entire classes of bugs, much like C prevents accidentally overwriting registers that you need the values of with side effects compared to handrolled ASM
>bunch of "cybernetic" nerds who think USSR could've avoided its disintegration had there been advanced computing system
Unreal stuff, lmao.
>>1860146its not just had that happened but whether or not the political will was there to make it happen. Bottom line is by the late USSR a bunch of the leadership already basically gave up on socialism anyway
>>1862248The Chinese bourgeoisie overthrew proletarian rule and chose capitalism with socdem characteristics.
>>1862248yeah, by liberalizing. but also it retains a planned sector as I understand
>>1862248china also sided with america and threw the soviets under the bus to save their asses.
> just graduated with a fucking bachelor's in computer engineering and I don't know what any of this shit means. what a stupid fucking waste of time and money
It's technical and specific domain engineer speech, if he spelles out the acronyms i think you'll somewhat understand
>>1864080Information security and defence at a free university.
High profile ones have the best people in the world that aren't working for a secret service.
it should clearly be run in a blockchain. no i dont mean a blockchain also used for currency. i mean one meant for
>guarantees of reliable results through parallel computation>not tons of it, but enough>arbitrary scalability through being able to join the blockchain and compute things very simplyits amazing that software engineers would trust some random C program.
https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_ReflectionsonTrustingTrust.pdf
>>1866509trusting trust is an issue for all languages, not just C
>>1866509>he hasn't written a symbolic debugger in assembly languageFilthy casual
>>1867207someone already posted this here:
>>1866470
>>1866509it should also be said that there is a process for bootstrapping gcc from an extremely simple hex monitor called hex0 that the user can write themself:
https://github.com/oriansj/bootstrap-seeds/blob/master/POSIX/x86/hex0_x86.hex0https://bootstrappable.org/projects.htmlhttps://bootstrappable.org/projects/mes.html>Since October 2019, Guix bootstraps by using MesCC—the small C compiler that comes with Mes—to build TinyCC, which is used to build GCC 2.95.0, which then builds GCC 4.7.4.https://savannah.nongnu.org/projects/stage0>A class of minimal bootstrap binaries that has a reproducible build on all platforms. Providing a verifiable base for defeating the trusting trust attack.stage1 has code that looks like this, with single-character labels (:3 and @5 below):
# ;; First pass function
# ;; Reads Tape_01 and creates our label table
# ;; Will Overwrite R0 R10 R11
# ;; Returns to Main function when done
# :first_pass 8a
:3
42100100 # FGETC ; Read a Char
# ;; Check for EOF
E000A0100000 # CMPSKIPI.GE R0 0
0D01001F # RET R15
# ;; Check for and deal with label (:)
E000A030003a # CMPSKIPI.NE R0 58
3C00 @5 # JUMP @storeLabel
# ;; Check for and deal with pointers to labels
# ;; Starting with (@)
E000A0300040 # CMPSKIPI.NE R0 64
3C00 @6 # JUMP @ThrowAwayPointer
the thing works itself up to being able to assemble a simple C compiler written in 4399 lines of assembly
maybe it's possible to make a C REPL instead. that might be shorter
btw, regarding the bounds checking discussion, tcc has such functionality built in:
https://bellard.org/tcc/tcc-doc.html#Bounds
>>1867329thinking about it, while it may seem like a purely academic concern, it should also be useful when bootstrapping new architectures. I see some riscv code in there for example
>>1867207thanks for posting this anon
>>1864080python and soldering iron
>>1881581>soldering ironmaybe learning about server hardware would be better anon.
People can imagine luxury space cybercommunist punk utopia easier than they can imagine the end of capitalism :))
>>1881861How would that not be the end of capitalism you insipid troll?
>>1882495Try getting another one of ur brain cells working and you can probably infer what I'm saying
>>1882992>>1881861I'm kindof with that anon. You can't really have cybercommunism (or any type of communism) without ending capitalism.
so… that doesn't make sense.
>ERP is either going to be C-like or Python
But why not a functional language? Feels like doing shit loads of maths while avoiding machine-operating code is what functional languages were designed for?
>>1883019Functional languages aren't good for numerical analysis. The point of functional languages is that they operate at a higher level of abstraction and make people more productive per line of code. They aren't supposed to be efficient in terms of execution speed or memory use. Look at the original classic languages where Lisp for example (functional/multi paradigm, dynamic, slow/memory hog, safer) sits on the opposite end of the of the spectrum from C/Fortran (fast, low level, efficient, unsafe). Functional languages like Scala are better for data analysis/engineering.
Functional languages almost by definition have no ability to pass by reference or implement algorithms that modify values in place. Look at clojure, a functional lisp on the JVM, even though due to being compiled to bytecode they have about the same execution speed (though some overhead makes clojure "slightly" slower) clojure ends up taking 2/3x more memory if not even more, despite the heroic efforts of the language creators.
If you want to use a functional language most languages like Haskell for example have a C interface (FFI). and use C/C++ libraries for the heavy numerical lifting.
>>1882992I want to see you actually justify one of your idiotic one liners for once.
>>1887804Very cool video, i liked the part where they discuss why market socialism will inevitable lead to capitalism in less than 10 years. That is extremely fast. A strong argument against socdems and market socialists.
>>1887875>In a closed economic system, money is conservedNice cryptozoology paper, dude
>>1889461it is though. money is created by accounting maneuvers that ultimately sum to zero
>>1889463The complements to any given entry in the ledger may not ever be realized, or be called upon to.
But it is true that economics is apologetics, not science.
>>1881861>>1882495No no, she's got a point.
People imagining a utopian future tend not to mentally bridge the gap between now and then.
>>1889465>The complements to any given entry in the ledger may not ever be realized, or be called upon totrue. inflation will surely make it so that the account credited when money is created is never debited in full. but that is also just surface appearance. every currency eventually goes through a period of revaluation where the old money is replaced by new money. such an operation
must zero out the account from which the old money was credited in the first place, either directly or indirectly. that is, it
de facto debits the account in full. pool's closed.
>>1889449thanks for posting anon
>>1889504<due to foreign aidsGood point. We haven't really had a redenomination in burgerland
yet, only the confederatebux writeoff and the general gold requisition.
>>1864080Computer Science & related fields would be an obvious choice. Second choice would be systems or industrial engineering, cybernetics. Third would be a more general degree math/applied math, physics, or even economics [as long as you read the coursework critically as part of bourgeois ideology].
Possible to create a decentralized app that coordinates workers, to achieve socialism in one fell swoop?
>>1892283the app isn't the hard part, the hard part is the buy in i.e. the politics.
>>1892313Why didn't Facebook need politics and buy-in though? As an app. Why does a socialism app need more?
Could a glitch in capitalism's logic be found that turns its economic incentives against itself to transform it from within? After all capitalism has internal contradictions. Can they be harnessed?
>>1852043Does anyone have any materials about soviet networking endeavours. I am not talking about OGAS. I know they already had some proto-internet projects that were implemented, like all-union databases and planning instruments. Materials in Russian are fine too.
>>1892283"decentralized" in what sense?
we certainly need software. but there is also a political angle. on the software side, the backend is far more important. any "app" you write is just a means of accessing the backend. we could imagine many ways tools of accessing it - phone apps, IoT devices, html forms, ssh
>>1892497the USSR started networking as early as 1959, see the Kossov interviews. not sure if the last thread got archived, so I'll repost them
>>1852217>unaccountable self serving bureaucratsReplace them with an algorithm.
>>1852897This works until you run out of a really important product like food and have to wait for the planning cycle to catch up. You need a system that can not only simulate supply/demand for the present planning period, but for future planning periods and adjust accordingly.
Realistically such a system would have to think faster than any human (think how difficult it was for the Soviet planners to do simple things like allocate goods trains). A hypereconomist continually trained and retrained on a wealth of human production/consumption data.
Highly undemocratic (you can't understand what the roboplanner is doing half the time, let alone vote on it), but on the plus side it would optimise for human needs instead of capital accumulation.
>>1892679What aboutt mentats instead?
Accountants on meth basically.
From birth.
I'm religious person, but I really trust socialistic full automation that It will end most of the problem we face to day, and get rid of capitalism problem which plague our believers.
>>1892698Human brains have high latency compared to silicon, so there's a hard limit on how fast they can work (even with unlimited supplies of meth). Also humans can be corrupted or coerced, which is the main problem with bureaucracy in the first place.
The best position for a mentat in cybercommunism is probably data scientist. The algorithm is only as good as the data you feed it, so someone will need to collect records of everything produced and consumed (with careful scrutiny so only real production is counted).
>>1892679>Highly undemocraticquite the opposite. fully automatic, fully open and fully centralized planning is the only way to democratize it
>you can't understand what the roboplanner is doing half the timethis is why you use linear planning and not dumb shit like """AI"""
abominable intelligence>>1892698>replacing blessed machines with disgusting flesh
>>1892901>replacing blessed machines with disgusting fleshI'm pretty sure human brains enter the composition of like half the Imperium's cogitators.
>>1892901How are you going to do that for an entire economy full of complex non-linear relationships?
>>1892921everything is linear locally. but more importantly, it's difficult to model non-linear things. an accurate linear model means the residue shrinks quadratically with respect to the planning period
>>1892908machine spirits aren't human. there is Imperium tech that straddles the line more extremely, like Thallax. some machines operate autonomously but aren't abominable because reasons
>>1892942How can everything be linear locally in a high-tech society? You're going to acquire production multipliers at an increasing rate as your local economy develops. Multiplication of multipliers = non-linear
>>1892997you probably think exponential growth is non-linear and yet
>d/dx exp(x) = xis a linear relationship
>>1893031>calculus existsWe're talking about countless products and services interacting and changing the supply/demand ratio of each other. How will you even begin to solve the resulting 138946th order differential?
>>1893038the Feldman model is locally linear for starters. even just looking at Marx' schema in vol II, which is entirely linear on a year-by-year basis, over time it is exponential. and an exponential is an infinite order polynomial
>We're talking about countless products and servicesthis just makes things harder for any attempt at non-linear operational planning. that said nothing says we
couldn't use a non-linear model, for example when modelling the environment. but unless you're proposing we use some kind of non-Lipschitz continuous thing, we can always discretize and linearize things. there is one big exception however: investment planning
>>1893050oh and shutting things down can also result in problems. discontinuities
>>1893050>the Feldman model is locally linearI understand a model can be locally linear, I just don't see how such a model could accurately reflect a real economy.
>we can always discretize and linearize thingsThis sounds like a lot of work for very uncertain results. Every time the real world gives you an exception to your model you have to develop a fix, which itself may have further unintended consequences.
Why can't we just keep the market economy, but replace rulers and owners with a network of competing deep learning algorithms? It would be like capitalism, but the workers own the capitalists.
>>1893077>keep the market economy>deep learning algorithmsI don't even know how to respond to this other than calling it incoherent nonsense
linear planning works well enough, we don't need to complicate things
>>1893077>Why can't we just keep the market economysee this
>>1887804 and
>>1887875
>>1893124>incoherent nonsenseCapitalist market economies perform well on productivity, but they produce inequality. So I'm thinking if they were worker-owned they would be good. If the capital allocation were done very efficiently by deep learning algorithms they would be ideal.
>linear planning works well enoughWhere has it worked really well? I mean beyond just subsistence level.
>>1893135>Where has it worked really well? The first 3 fiver year plans of the soviet union, as an example?
>>1893139Where did it go wrong after that then? Did they adopt a bad plan or was it unforeseen circumstances?
>>1892283>>1892657Decentralizing is a meme. Computation must be centralized.
>>1892679>You need a system that can not only simulate supply/demand for the present planning period, but for future planning periods and adjust accordingly.Sure. Even short term reactions should not really be reacting to do what is at a point in time, but rather look at the change over a short slice of time and extrapolate that a bit and react to the next moment according to forecast rather than the current moment. (Any child that tries to hit a moving object with a ball is already doing that.)
>Realistically such a system would have to think faster than any humanAny pocket calculator thinks faster than a human about some issues :P
>(think how difficult it was for the Soviet planners to do simple things like allocate goods trains)Do you mean a generalized traveling-salesman problem or what? Not simple.
>Highly undemocratic (you can't understand what the roboplanner is doing half the time, let alone vote on it)The complexity of judging whether something works or not is often much lower than the complexity of making that thing. You can judge the realism of paintings at a very high conceptual level and making all sorts of distinctions, even if your own hands fail you when it comes to painting things. Why would this be different.
>>1892942>everything is linear locally.Also, some non-linearity is fake news: When there are things that are distinct in quality and people have local knowledge about it and choice (like units of some tool with different levels of wear and tear) while the center has a formal representation where the things appear as equal, there will then be the appearance of a diminishing marginal effectiveness, which would disappear under a more fine-grained classification.
>>1893349>Decentralizing is a meme. Computation must be centralized.Maybe justify this? Am I going to trust some random computation and believe whoever tells me the results? No, I want it to be reproducible which requires decentralization. If it's too complex to compute on consumer hardware then break it down where I can verify some core segment within it. Decentralizing is a meme? You sound really ignorant especially since you followed up with nothing. Are you an engineer?
>>1893349>Decentralizing is a meme. Computation must be centralized.agreed. those who advocate for "decentralized" planning need to be very clear what exactly they want decentralized and why
>there will then be the appearance of a diminishing marginal effectiveness, which would disappear under a more fine-grained classificationcould you elaborate on this? for some processes we'd only have statistics to go on, and doing much better than an approximate linear model would be difficult. even with say chemical processes that are well-studied, there's always some uncertainty. some parts of the economy aren't even processes at all, but mere resource sinks, such as hospitals
the planning system can only really provide suggestions. ultimately the plan has to be concretized into actual orders. here there may be some leeway. workers have the local knowledge to see things that the planning system does not see. but if the information given is accurate enough, workers only have to deal with the "slop" and not the meat and potatoes of large industrial processes without which society would come to a halt
>>1893375>Am I going to trust some random computation and believe whoever tells me the results? No, I want it to be reproducible which requires decentralizationno, if you want plan computations to be reproducible you want them to be fully centralized. it is only by completely centralizing all data, all constraints, all models etc that you can even hope to reproduce the results. once fully centralized, it is also trivial to ensure the data are mirrored in multiple location. for example of such centralized systems, see git and IPFS. one can have multiple datacenters that synchronize data in a hypercube grid ála MPI_AllGather such that everyone has the same shared state. this can be verified by sending Merkle DAG hashes around. that way everyone is struggling over the one
and only one set of data, fully centralized and visible to all
>>1893135Performs well on productivity? As time goes, the productivity goes up, but this means labor is reducing and thus less profit is made. Due to low profit, productivity upgrades might be delayed.
How can this be solved by allocation?
I see solution by letting people work where they want, so they do not wait for someone/something to allocate profit to hire them to do the upgrades.
>>1893573>some parts of the economy aren't even processes at all, but mere resource sinks, such as hospitalsAnd management. Don't forget ALL of management.
>>1893626If hospitals, the communal institutions concerned with the reproduction of actual life, are mere resource sinks, then there is no reproductive institution that isn't.
Unless our propellerhead up there is thinking of an alternative mode of medical provision and trauma care, which doesn't seem to be within his cognitive ambit or world model. Tbh he talks like he works for an NHS privatization contractor.
>>1893229I'm not debating it, I'm literally asking what went wrong. Thanks for the PDFs.
>>1893349>Do you mean a generalized traveling-salesman problem or what?I just meant setting the correct quotas for transportation such that goods arrive on time with minimal waste. Even if you had the optimal routes you would still have to allocate the correct number of train journeys to each line for a given time period.
>>1893375>I want it to be reproducible>If it's too complex to compute on consumer hardware then break it down where I can verify some core segmentBetter yet, run the whole thing on a blockchain and shard it. Everyone can collectively contribute to the planning algorithm.
>>1893519That's just one idea. The AI in question would be collectively owned by the workers so they keep 100% of the profits.
>>1893573>those who advocate for "decentralized" planning need to be very clear what exactly they want decentralized and whySome of the best machine learning systems use an adversarial model in which multiple algorithms (or multiple instances of the same algorithm) improve each other with iterative feedback. This is very similar to the way markets use the feedback loop of competition to make prices reflect supply/demand. If you could simulate all the nodes of a market on an adversarial neural network you could reproduce the productivity results without needing capitalists/bureaucrats/[insert ruling class here] as resource allocators.
>>1893610Markets get around it with leverage, so future profit can be invested in the present. Why can't a communist planning algorithm do something similar? Allocate future profit to a present upgrade.
>>1893647Since you're managing capitalist properties and capitalist relations anyway, just nationalize Salesforce and have done with it
>>1893623overhead in general. while a sulfuric acid plant will mostly consume sulfur, at a ratio of one ton of sulfur per three tons of acid, it will also consume office supplies, coffee, birthday cake for Kevin and so on in addition to labour. the labour itself could be divided up into the people running the actual plant, accountants, jannies and so on, but also people whose usefulness to the workplace is dubious. so long as the embodied labour of the output of the plant is roughly the same as every other sulfuric acid plant then who cares
with reproductive services things are different. there are no "knobs" to turn with which to adjust output. instead you have concerns like the number of hospital beds and the number of patients per nurse. such measurables need to be kept within certain bounds to provide a certain level of service. the cost of said service can simply be socialized via tax, or we could choose to charge patient fees, or both. nothing says we wouldn't tax fat people or alcoholics whose unhealthy lifestyles are a burden on society
>>1893703>the cost of such serviceNo, no, that's just capitalist management i.e. idealism. Materialist management deals in I/O tables and material, not "value". If you didn't know that then you should lurk while reading Cockshott.
>>1893710anon, you do realize that the various goods that a hospital needs all has embodied labour associated with them, and that changes to the demand thereof may lead to changes in the division of labour? of course these goods also have to be tracked, and they will
>I/O tablesI'd like to see you try to describe the production function for a hospital, as opposed to say always giving all hospitals what they want, or assigning a budget, or some combination of the two (take as many gowns as you need, but MRI machines are still scarce and need to be rationed)
>materialist management does not deal in valueau contraire! value (SNLT) is transhistorical and we should expect it to be managed
far more intensely in a socializing/socialized economy, labour being the ultimate input and something that is not to be squandered. perhaps in higher-phase communism we might not need to, but that is hundreds of years away at best and hardly worth worrying about
>>1893647>Some of the best machine learning systems use an adversarial model in which multiple algorithms (or multiple instances of the same algorithm) improve each other with iterative feedback. This is very similar to the way markets use the feedback loop of competition to make prices reflect supply/demand. If you could simulate all the nodes of a market on an adversarial neural network you could reproduce the productivity results without needing capitalists/bureaucrats/[insert ruling class here] as resource allocators.the next AI winter can't come soon enough..
>Why can't a communist planning algorithm do something similar? Allocate future profit to a present upgrade.this is precisely what the Feldman model is about, minus the profits of course
>>1893659Salesforce still needs capitalists and bureaucrats to operate it. I'm talking about building a system that can automate those jobs entirely so that 100% of profit goes to the workers.
>>1893725>value (SNLT) is transhistoricalThen capital itself is transhistorical, and you have completely lost the plot.
>>1893768What is profit from a marxist analysis
>>1893776>>1893789The difference between input and output. Currently scraped off and reallocated by capitalists. They want to automate work so they can take the lot, but if workers automated them first the workers could take it all back.
>>1893792>eternalizing private propertyRetard
>>1893788>Then capital itself is transhistoricalno. SNLT exists in all modes of production, as does use-value. exchange-value does not. or maybe you suggest that labour doesn't have a social character in say primitive communism? that a person could get by while shirking?
what you should disagree with is that value == SNLT. but then we have the problem of figuring out what value
is>>1893792>They want to automate work so they can take the lotyou didn't read Capital, I can tell
All of you Cockshott people are smelling your own farts and have gone insane. The entire thing rests on a mystification of the computer and human gone apeshit. So much fucking eugenic creed in this hellworld.
>>1893725Value exists because there are humans or entities capable of knowledge and reason that can value anything. By definition it arises in history, referring to a world that is presumed to have existed before anyone valued any part of it.
All of this comes back to humans choosing to make other humans suffer. That is the only thing that really is valued - the use of some thing for making another human suffer. If we wanted nice things, we would have to work with objects in-kind for what the are or what they can be, by some system we devise for that purpose. The economic task in human society exists entirely because there are armies of men who will make sure you starve, for the sake of ensuring that you starve. There is no way to make that into something it is not.
Put another way - if there were no struggle for survival in the world as the result of humans attacking other humans, nothing would be "valuable" in the sense that political economy regards such a thing. Something might be valuable as a food source or energy source, but those things are not freely exchangeable with each other, and the qualities of different foods or fuel for energy are not freely exchangeable or comparable at all. It is only possible to compare them by developing some general theory of food, or energy, or matter, and so on.
But, you can have all of the food and energy in the world, and if the will of the human race is to make each other suffer like this, the result will always be the same. If the will of the human race is not to make each other suffer, then the management of resources is a trivial task in any era, so trivial that it would be beneath our notice and our dignity to wage wars over it. No functional battle formation or family would engage in this incessant intercine struggle for struggle's sake. That has always been a way to immiserate the rank and file so that they fear officers and sacrifice, while the favored classes are granted absolute impunity and no consequences whatsoever for it. It's a way for families to designate who the stupid ones among them are, to feel better about themselves.
The moment the eugenist insinuations begin, I want nothing to do with this Satanic race - so much, that acceeding to it for a moment is so ghastly that I will spend enormous sums purely to never have anything to do with fucking Satanics. Exterminating a Satanic race is no loss to me.
This, of course, is what the eugenic creed wants - for Satan to win, and cull with maximal torture anyone who was honest, or wanted a world other than this shit.
>>1893991>>1893989>>1893987who the fuck is this schizo
>>1893816Fair enough on SNLT. However I am more certain that use-value exists outside of production.
>what you should disagree with is that value == SNLT. but then we have the problem of figuring out what value isYes, exactly. Something something the importance of actions but it takes 1500 more pages to describe how to make them move and articulate. However I've been meaning to check out Wertkritik…
>>1893995Nta, ftr
>>1893573>doing much better than an approximate linear modelThat's not the point. The point is: A situation's formal representation can look like using an input with diminishing returns, when the formal representation groups together distinct things with different efficiencies. So the approximate model is NOT linear when changing the output scale, and a model with more detail (now distinguishing several alternative inputs) might actually be linear for varying the output for each input as an isolated pure process. People tend to use the more efficient input when they have a choice, whether that is represented in the model or not, so under the old model it looks like somehow "the input" gets less and less efficient at bigger scale. You have this effect everywhere where employee efficiency is crudely measured by units that treat as equal what is not equal and what is easily recognized by people on the shop floor as not equal.
Consider people working in a warehouse, stowing things from carts. Of course, some carts are marked and most be stowed before other carts, but there is no full strict ranking of all stowing carts that you must follow. Suppose there is some crude efficiency measure for the workers and they know about it. As long as an employee has some control over which cart to take, they can game the system. If the measure is number of items, take a cart with lots of small items on it. If the measure is number of carts emptied, take a cart with few items. In either case, the vulnerability of the measure comes from treating different things as the same in the formal representation. The formal representation will show a drastic fall in marginal efficiency of workers. Of course with adding workers there is a point when people start to step on each others toes, but the picture of falling efficiency is exaggerated. Each cart has to be stowed by somebody. The fewer carts there are waiting, the less workers can game the system, so of course the newly added workers will tend to appear as less efficient.
>>1893634You are stuck mentally just because of what you read into an expression you are not familiar with. Imagine the other poster is not at all in favor of reducing hospital care or privatizing it or whatever, read the post again, and try to figure out what could be meant by the expression. Assume the expression has no negative association for that person.
>>1893984absolutely subjective
>>1894015>I am more certain that use-value exists outside of productionduh, of course use-values do. all of nature is full of use-values that reproduce themselves without
much human interference
>Something something the importance of actions but it takes 1500 more pages to describe how to make them move and articulateI would suggest staying away from value criticism because you're just going to tie yourselves into philosophical knots compared to just axiomatically asserting value == SNLT. the best I've managed to get out of value critics that isn't endless yapping is that value is a social thing. but that's already covered by SNLT
>>1894364>So the approximate model is NOT linear when changing the output scaleyou can do diminishing returns with a linear model. rising marginal costs imply convexity. the real challenge is bringing new production techniques to bear, which tend to lower costs. figuring out when to introduce such changes is non-convex programming, which is very difficult, because it is fundamenally combinatorial
>a model with more detail (now distinguishing several alternative inputs) might actually be linear for varying the output for each input as an isolated pure processyou mean joint production? that's easily handled by LP
>People tend to use the more efficient input when they have a choice, whether that is represented in the model or not, so under the old model it looks like somehow "the input" gets less and less efficient at bigger scaleof course. given a certain technical makeup, production tends to be convex
>As long as an employee has some control over which cart to take, they can game the systemthis is the kind of problems the Soviets ran into with aggregate planning. they had to come up with plan indicators, which were often possible to game. add a bonus system into the mix and you have more problems. hence why we advocate for disaggregated planning
>>1894463>new production techniques>you mean joint productionNo I am actually not talking about every issue under the sun simultaneously. In post
>>1893349 in the section that starts with
some non-linearity is fake news I introduce one specific issue. In post
>>1894364 I gave an explicit example for what I mean. Just go through the example and distinguish between what is actually happening on the shop floor and how it looks in the formal representation.
>>1894476>Just go through the example and distinguish between what is actually happening on the shop floor and how it looks in the formal representation.the map isn't the terrain, yeah? this is well known
>>1894505It's about the
type of the map-territory mismatch. And that
this is not well known to you is proven by your own comments.
Cheers.
>>1893647>>1893768>by the workers so they keep 100% of the profits.>jobs entirely so that 100% of profit goes to the workers.First part of the video is a basic raw explanation of the theory of labor of marx. Basically, if all value is the product of labor, then having wages equal to value produced would result in not having value left to expand business. The minimal value produced must be = wages + cost of maintenance + necessary expenses + surplus.
Lets say, in a hypothetical future socialist society, all workers would receive, instead of 100% of all profit made in state own industry and farms, cooperatives, etc, they received 75% of all profit, while the rest would be used both to increase production, buy more machinery, invest in innovation, and be collected by the dictatorship of the proletariat, so that the the main government can redistribute its value to pay for schools, hospitals, armies, etc. Those things don't create "value", but they need to be payed somehow.
Basically, is not enough to reunite the worker with the value produced, we must completely destroy the working class as we know it and destroy the origins of exploitation, the paradigm of value/production. Capital arises from the competition of goods to one another. What Marx proposes, instead, is communism.
topics include the theory of value, commodity fetishism and definitions about what is value, commodities and prices.
>>1894536If the things you are reinvesting in belong to the workers then 100% of the profit is still going to the workers. It's just a question of how much you allocate to present consumption vs future productivity. Not a problem.
>state own industry and farms, cooperatives, etcWhat troubles me is you envision a "state" as separate from cooperatives. Historically this has not gone well.
Can we agree to experiment on this cyberutopianism after we establish market socialist economic model?
>>1894599No. Die in a house fire, think tankie
>>1894600Its unnecessary to go the hard way like with the great leap forward failing and leadership changing.
>>1894529Not scared. Besides, SNLT has the problem of eternalizing labor, opposed to Marx's big idea to critique political economy and its categories, including value itself, goals which the revisionists have thrown aside along their way to the genteel middle class.
>>1894539This guy knows what's up. And with that there will need to be made leading changes to the systems and processes of production: the further development of lights-out manufacturing, a general system of labor dispatch fit for it, a social order compatible with it… Anything less serves only in perpetuating industrial capitalism — you're not in "lower phase" until you start running factories under non-value regimes e.g. such as a simple necessary/surplus priority model.
It is telling that this thread is filled with managers and accountants. Notably, not a single industrial or robotics engineer in the house, basically no discussion about how a properly communist MoP might look and work. Not a thought about what social surplus might look like to the individuals who have their own purposes at any given moment.
Only GOP unionoids like >1894605 trying to save his boat.
Sad.
>>1894607>categories, including value itself, goals which the revisionists have thrown aside Its called science. Theories must adjust to empirical evidence. Nobody is demanding dogmatism from you like when USSR existed.
>>1894619>it's called scienceChopping off the part of the phrase that scares you
>critique political economy and itsisn't very scientific, is it? Why, isn't that exactly that same inversion Kautsky and Lenin asked of us, toward the particular task of reconstructing a feudal society into a properly capitalist one, a task that has been completed in world history and toward which any further effort only takes us further from the end of class, money, and the state?
>>1894639Its great to have ultra retards with us but I wouldnt call Lenin or Mao dogmatics. They innovated and expanded theory.
>>1894650>ItsUnderage
>innovated and expanded theoryOnly in the backward direction, and in every case these theoretical expansions merely elevated the contingent and conditional to eternal verities. First Kautsky's social democracy then Lenin's "Marxism" then "Marxism-Leninism", all built on the principle of upgrading feudal conditions to pure capitalism
in order to get past it. A historical necessity in their times and places, but no longer.
>centrism is the truthRightoid trolls don't even try to hide it anymore
>>1894607>SNLT has the problem of eternalizing laborSNLT will always exist. do you think people wouldn't work in higher-phase communism? that managing material resources wouldn't become more important? almost certainly shirking wouldn't be acceptable. you'd still need to figure out where people are needed, and you'd still need to fucking account for things, include labour, as well as the types of labour at the system's disposal, and whether training more workers of a particular skill is necessary
>muh managers and accountantsaccounting will become
more important in communism, not less. see vol III of Capital.
>robots and shieetI work with embedded systems and I can tell you that full automation will never happen. you can go have a look at any roboticized production line and you will - surprise surprise - find people there, overlooking things, repairing things, tweaking the code and so on. I have a relative who works with designing production lines, and we frequently talk about some of the problems associated with it
>>1894687>the USSR was le capitalistand there it is
>>1894687Congrats on making other ultras feel like revisionists.
>>1894721>do you think people wouldn't work in higher-phase communismIf they are still "working" in a "workplace" over which they have only formal ownership and not real control, for a wage just adequate to reproduce them as subordinates trapped in some Other's dream, it's still capitalism. Obviously your vision is an extension of your own condition.
>accounting will become more important in communism, not less. see vol III of Capital.You're right, but that's not what I said. Bad faith to transsubstantiate the worker into the work, as if the specialist were required. III.17, Commercial Profit:
<The commercial worker, in the strict sense of the term, belongs to the better-paid class of wage workers — to those whose labour is classed as skilled and stands above average labour. Yet the wage tends to fall, even in relation to average labour, with the advance of the capitalist mode of production. This is due partly to the division of labour in the office, implying a one-sided development of the labour capacity, the cost of which does not fall entirely on the capitalist, since the labourer's skill develops by itself through the exercise of his function, and all the more rapidly as division of labour makes it more one-sided. Secondly, because the necessary training, knowledge of commercial practices, languages, etc., is more and more rapidly, easily, universally and cheaply reproduced with the progress of science and public education the more the capitalist mode of production directs teaching methods, etc., towards practical purposes. The universality of public education makes it possible to recruit such labourers from classes that formerly had no access to such trades and were accustomed to a lower standard of living. Moreover, this increases supply, and hence competition. With few exceptions, the labour power of these people is therefore devaluated with the progress of capitalist production. Their wage falls, while their labour capacity increases. 3 The capitalist increases the number of these labourers whenever he has more value and profits to realise. The increase of this labour is always a result, never a cause of more surplus value. (a)Engels' footnote a:
<How well this forecast of the fate of the commercial proletariat, written in 1865, has stood the test of time can be corroborated by hundreds of German clerks, who are trained in all commercial operations and acquainted with three or four languages, and offer their services in vain in London City at 25 shillings per week, which is far below the wages of a skilled fitter. A blank of two pages in the manuscript indicates that this point was to have been treated at greater length. For the rest, we refer the reader to Book II (Kap. VI, S. 105-13) ("The Costs of Circulation"), where various matters belonging under this head have already been discussed.— F. E.Grundrisse chimes in with an important point:
<However, take the case of the lithographing machine, which everyone can use to make lithographs without special skill; suppose the value of the instrument immediately upon its invention to be greater than that which 4 workers absorbed before these handy things were invented; it now requires only 2 workers (here, as with many instrument-like machines, no further division of labour takes place; instead, the qualitative division disappears)Marx again:
<All these costs are not incurred in producing the use value of commodities, but in realising their value. They are pure costs of circulation. They do not enter into the immediate process of production, but since they are part of the process of circulation they are also part of the total process of reproduction.
>I work with embeddedSo you're deeply invested in your material conditions: access to a general servant class that responds to your commands, refrains from punching you when you act like an ass, and is willing to perform services for you that they can't afford for themselves. That explains why you're so worried about skivers, too.
As for me, I can afford to dine out about once a month. I have no reason to respect your social position. Perhaps you are the intellectual bourgeoisie here and the communist inversion will not be so kind to you and your barracks utopia as you hope.
Let's imagine a very, very, very small role for markets. What I'm describing in the following is not what is usually meant by "market socialism", but something far more constrained in its use of markets.
Suppose the workers control collectively the machine of production. They collectively set in advance the quantities to be produced, the prices, the salaries. Suppose they plan to produce enough of item X to give everyone one unit, and they actually do produce that much then and they set the prices and the salaries in a way that item X is something everybody can afford it and wants to at that price.
Question: Does it follow that everybody actually gets the item?
Answer: Coordinating well in advance the quantities, prices, salaries improves the odds; but does not make it certain, because some individuals may obtain more than one unit, so some will have to go without. This is how chaotic the market is, even with all these regulations in place about produced quantities, prices, salaries. Just about everything is set in place an arranged in the right way, there are the things to consume in storage facing the people who want these things, and there is just this tiny gap between the things and the people and the people use play money to bridge the gap, and the play money can be designed electronically to be impossible to switch hands between people (the consumption points just get deleted from the system when obtaining the consumer item) and the play money can be designed to have an expiration date as well, and this stupid shit still happens.
>>1894599sure, even cockshott says workers coop economy will have to be a necessary step before cybercom
>>1894867Stop, stop. You Cybersoc idiots are killing my brain.
A market entails agency of its participants. If you're suggesting a market where everything is controlled and predicted, you're not operating with a market, and this would be evident in how participants view the market.
You don't need to make it a hypothetical because this is effectively what happened - the manipulations of finance are documented well enough and expected, and there isn't a great secret about how things are produced.
That has been the problem with "cybersoc" - assuming the market logic is natural and operative outside of agency, when the entire reason a market exists is because of agents wanting particular things, rather than arbitrary and infinite things. This might come as a surprise if someone huffs ideology, but the wants of market participants are not infinite and arbitrary. The coincidence of wants that makes a market possible arose because among those wants are security, liberty (at least enough to exist at all and not be choked by hostile and alien managers obsessed with managing your life), and an answer to the fear of that ever-present threat to humanity - other humans. You can solve that problem through cybernetic regulation to a large extent, but the starting point is that you would do away entirely with dickering over who is entitled to what miserly share of the social product. It would simply not be a question regarding basic necessities, and opulence and luxury would be just that - a marker of petty distinction, or something allocated sparingly for social ends, usually involving things which would be disinteresting to most people as property of their own.
If you wanted to regulate human social behavior with cybernetic planning, you are asking very different questions. The economic problem is very trivial if it is about resources. Regarding the necessities and reasonable wants of people in society, that is exactly what the problem is - we have food product that is measured out specifically to starve people because it could keep them at the edge of misery and make them pliable to go along with things they really don't want to do. There are reasonable expectations of security that were once upon a time so expected that autocannibalizing a society would seem like madness, outside of eugenism.
That was really the basic proposition of socialism - not micromanagement of private life, but answering the most basic needs by a rational plan, which was well within the capabilities of a bureaucracy in the 19th century. It was not a technological problem, as if people were too stupid to allocate numbers on a spreadsheet with sufficient information about who asked for what, and what their genuine needs were. Eugenism has been entirely about distorting those needs and wants and morphing them into pure tools of eugenism. That becomes the sole remaining value, and why it appears the economic problem is "intractable" - because the agents value making humans suffer more than the thing simple social purported to do, and could easily have done. It was so easy that in some ways, it happened in spite of the prevalent thinking, simply because not dickering over misery is a lot more efficient than finding new avenues of smugness and contempt. It got bad enough that capitalism had to focus on baubles, fetishism, and pornifying the whole society to make their stupid religion temporarily acceptable. Now it drowns in rot and filth for the stupidest and faggiest purposes.
Before you get into grand schemes to micromanage private life, there are some basic utilities and services like, I dunno, food - food which the USSR never could socialize, which was understood to be a failure of the project. We could do that now, except the dominant ideas are about engineering famines because that would keep the people fearful and respect these technocrats gone apeshit.
Also, energy, which is already controlled by monopolies and centrally planned because that has been the basis for this empire's transportation, motor force, command of information. Somehow the existence of this very large interest in energy is invisibilized and we pretend the world didn't change from a toy model in the 19th century simulation, except the 19th century wasn't that either.
The point is, you'd have to change a lot of things to make this cybernetic planning work as a type of socialism… except many of those changes are happening because cybernetics is too obviously useful, if not for the insane fetish about information eugenism imposed on the world purely to make us suffer for their faggotry.
>>1894867>Here is why markets are shit in plain language.>>1894906>you're suggesting a marketI'm suggesting that you have brain damage.
>>1894815boy you sure have me figured. how dare I care about practical concerns like stopping global warming, a task that will require more sophisticated coordination than a war economy? silly me
>>1894867>play moneyCounterproposal: Fractional Labour Voucher Blockchain Direct Democracy
Create a work schedule for a given time period detailing which days you are willing to work, what kind of work you can do, the range of locations you can travel to and any conditions or requirements your clients must meet. Turn this into a smart contract and exchange units of labour for consumer goods and services. A central exchange allows people to trade the labour they have been paid for the labour they need. The bearer of your labour may thus trade it for someone else's labour or redeem the smart contract and summon you to work for them.
After labour is performed the worker and client anonymously rate each other. Everyone therefore ends up with two ratings attached to their name - a worker rating and a client rating. Smart contracts can include rating as a condition for employment, such that persons with a low client rating cannot buy any fraction of the contract, protecting everyone from bad working conditions. A person's worker rating is included on all smart contracts they create and is visible to potential buyers, ensuring that the market price of smart contracts accurately reflects the quality of the labour they represent. (Note this also makes labour fully fungible, as any two hours of labour with the same rating have the same value, making them interchangeable on the exchange.)
Every individual is given a blockchain identity at birth, and cooperatives may create collective identities to exchange their group labour. Capital goods are collectively owned and must be rented by individuals or cooperatives in exchange for labour. Tenancy is recorded to the blockchain so there can be no dispute over rightful possession. The rates are set automatically by control loops to ride the peak of the Laffer Curve, such that rents provide maximum revenue. This generates a pool of collectively owned labour which may then be spent on public goods and services.
Each individual may propose either one public spending project or the abolition of one existing project per planning cycle. Anyone may rate any proposal on the blockchain. At the beginning of each new planning cycle the highest rated proposals are automatically implemented. Each individual may then propose a budget to allocate public revenue to projects. The average of all the proposed budgets becomes the public budget for the planning cycle.
A similar process is used to elect public officials, with individuals rating each other and then voting on which of the top-rated people should run each project. The only legal tender is the labour-backed smart contract and nothing else may be traded on the exchange or paid as site rent. If trade with foreign nations is desired it can be conducted in gold or silver.
>>1895038>Fractional Labour Voucher Blockchain Direct smart contract peepee poopooNo.
And the solution was already alluded to in the last thread :P
>>1895038>bockchainJust use a normal database
>>1895068Blockchains are immune to tampering by corrupt bureaucrats. They can also be implemented on a minimal budget, because the compute is spread across all personal devices instead of being done on central servers. You don't even have to wait for the revolution - you could make a labour voucher blockchain right now and get coops using it to trade with each other.
>>1895076you could use something like git or IPFS, which technically are blockchains but few people refer to them as such
>>1895076>>1895083people really need to study distributed systems because they get distributed systems and blockchain mixed up. There are tons of possible distributed and even peer to peer styled systems that don't involve blockchain or hairbrained proof of work schemes.
>>1892818the data scientist is probably the cybercom equivalent of an economist in that they're an analyst. The actual gathering of information is done at the point of production so that job would be called like industrial observability engineering or something.
>>1895088>There are tons of possible distributed and even peer to peer styled systems that don't involve blockchainHow do they prevent tampering or DDoS?
>>1893573>no, if you want plan computations to be reproducible you want them to be fully centralized. it is only by completely centralizing all data, all constraints, all models etc that you can even hope to reproduce the results. once fully centralized, it is also trivial to ensure the data are mirrored in multiple location. for example of such centralized systems, see git and IPFS. one can have multiple datacenters that synchronize data in a hypercube grid ála MPI_AllGather such that everyone has the same shared state. this can be verified by sending Merkle DAG hashes around. that way everyone is struggling over the one and only one set of data, fully centralized and visible to alluhhh, okay then we are in agreement, except guess what you just described a decentralized system. not centralized..
>>1893647>Better yet, run the whole thing on a blockchain and shard it. Everyone can collectively contribute to the planning algorithm.very based
>>1895319A system of computation doesn't change from centralized to decentralized by adding backups.
>>1894867markets will likely have a role in a transitional period, quite a substantial one even
we should expect companies, even cooperative ones, to purchase things from the planned sector. arguably one task that lies before us is to take socialist commodity production far further than the USSR ever did. the PRC does something like that, but not far enough
>>1894906>A market entails agency of its participantsimagine believing this
>>1894911>well within the capabilities of a bureaucracy in the 19th centurylol
>It was not a technological problem, as if people were too stupid to allocate numbers on a spreadsheet with sufficient information about who asked for whatlmao
the implication here is that the Soviets were retards
>>1894916>food - food which the USSR never could socialize,<what are sovkhozes
>>1895121>How do they prevent tampering or DDoS?there's a million ways. signed hashes prevent tampering. DDoS can be dealt with by using VPNs
>>1895319>what you just described a decentralized system. not centralizedno. for an example of a decentralized system, look at DNS. there is no single global state with DNS. with planning we
want a single global shared state. all data must flow into a common "cloud". or a git repository. or whatever.
Is it possible to do this like in 2D? Like, and EMP fries the lecky so we need to go back to the 2D.
>>1895380Imagine believing that people engage in market activity for some reason unknown to them, as if they randomly walked into a market and randomly exchanged tokens in this particular way without knowing why. You realize how retarded that is?
Markets only exist in conditions where they are sensical to the participants. A warlord in a society without laws is not operating a market and has no need of a "market". Markets are destabilizing and uncertain in such conditions.
If the participants in a market knew the outcome of all exchanges, they could and would have every reason to bypass the market mechanism completely. That is what we do, since most human behavior exists outside of the market entirely. Otherwise, the most basic mechanisms of life would be doled out by market activity and haggled over incessantly. That's what happens in a market. If the market is solved, there is no purpose in participating, and the members would already arrive at the outcome before any exchange took place. That already happens in a market, because the participants are aware of their possessions and objectives before the market formed.
This is all very, very basic to what a market is, and why markets were an unusual thing rather than default. Where markets exist, they were local and temporary before they were established as permanent, then world spanning markets. On this, every economist before the 20th century agreed, and the non-retarded economists of the 20th century and since concurred. It's only in the retarded fag ideology that it was naturalized and made essential.
In other words, the market wouldn't exist if it were truly uncertain and random. The market is not there to solve an information problem, and never was. It's there because humans were antagonistic towards each other, and the market was the chosen mechanism for promoting internal conflict and the preference of interests in society.
So long as the market exists, you're still supposing the antagonism exists. That's why a market exists. Anything cooperative from it occurs in spite of the market rather than because of it.
We could of course live with antagonism, but you can't make that into something regulated and "natural" without naturalizing the antagonism and making it permanent. That's why you people are insane - not because you'd make it permanent, but because you actually think reality is controlled by these smug assertions and such faggotry.
>>1895514>If the participants in a market knew the outcome of all exchanges, they could and would have every reason to bypass the market mechanism completely.>If the market is solved, there is no purpose in participatingWhen a market is even remotely close to being solved a period of creative destruction will occur as participants try to unsolve it for profit. The closer it gets to the solved state the more profit there will be in generating new products and services to unsolve it.
>A warlord in a society without laws is not operating a market and has no need of a "market".Hence capitalism is more productive than warlords or other feudal arrangements, as it can constantly generate more creative destruction. If you want to build a better mode of production you need to learn from the achievements of the current one.
>>1895517>The market is not there to solve an information problemTeleologically it is not "there" to solve anything - people don't design markets, they tax and regulate them post facto. Markets do amalgamate vast amounts of information, but of course the profit lies in the correction of information and therefore could not occur in the presence of perfect information. If you really want something better than markets and profit you need to design better information aggregators.
>>1895385right. Planning has to converge and doing it in a single computer or set of computers is way faster than adding the network latency of a whole country or even planet
>>1895485If an EMP fries society you have bigger problems. Any society without electricity basically reverts to amish levels of technology and subsistence farming in which case you don't need complex planning anyway
>>1895708>right. Planning has to converge and doing it in a single computer or set of computers is way faster than adding the network latency of a whole country or even planetA lot of posts here are really focused on the socialism stage before communism. If it's centralized there is a central authority controlling planning, or in other words a state. It must be decentralized for there to be no state.
>>1895752>It must be decentralized for there to be no state.thats not really true. You are confusing a government and a state. The state is an organ of class rule. It consists of special bodies of armed men having prisons, etc., at their command. You can have a stateless society that has a government. A government which is a consequence of self acting organization of a classless society is not a state. The state must have repressive powers the purpose of which is to reconcile the social antagonisms of a class society. Once economic and material equality have been produced and class has been abolished there will no longer be a need for a state as such, but that does not obviate the need for organization. With the end of class, the state will wither away. But this does not imply the withering away of any form of organization at all.
Nor does it really prevent centralization of planning in a political, let alone technical sense. Whether or not to perform a computation on a certain computer is an engineering decision, not a political one. It's perfectly possible for a computation to be performed centrally in a technical sense while being "decentral" i.e. "democratic" in a political one.
>>1895752"Lower stage" is basically the dictatorship of the PMC.
>>1895769>if we pretend the PMC isn't a class, we can pretend their dictatorship isn't a state
>>1895769>Nor does it really prevent centralization of planning in a political, let alone technical sense. Whether or not to perform a computation on a certain computer is an engineering decision, not a political one. It's perfectly possible for a computation to be performed centrally in a technical sense while being "decentral" i.e. "democratic" in a political one.you sound like you might be right but i think youre missing key technical points. a centralized system can have backdoors without anyone knowing, very complex ones
https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_ReflectionsonTrustingTrust.pdfdecentralized is going to have a much harder time suffering from such a problem since we can assume with socialism and communism we will have more variety of hardware and software and redundant decentralized software components written in completely unrelated languages and architectures, for safety
>>1896494To continue, in other words, it can be coopted from within or commandeered much more easily. Even if we trust this "non-state" governmental group that controls planning of all production, from a sociopolitical standpoint (lol sorry but "not a state" my ass) then let's think technically, can they technically control the system even with good intent?
the answer is no, and for that you'd need a decentralized system of the sort i described. and dont think this is minor quibbling, it's completely doable especially with blockchains, and an important architectural point to get right from the very beginning. or else how do you migrate later? you wont.
>>1896494that research paper is completely unrelated to anything being discussed.
>>1896578it's a clear vulnerability a centralized system is subject to, one of very very many. im sure you can think of more issues if you tried.
>>1895708>right. Planning has to converge and doing it in a single computer or set of computers is way faster than adding the network latency of a whole country or even planetthat is true, but that's also not my point. yall mfers don't seem to understand that logical centralization isn't the same thing as physical centralization. like this guy
>>1895752>It must be decentralized for there to be no state.petty bourgeois nonsense. this guy
>>1895769 is right. state != government
>>1896494https://savannah.nongnu.org/projects/stage0/full centralization is a necessary but not sufficient condition for all phases of communism. without it you're going to have what amounts to corporate secrecy, and you're going to have exchange. the Soviet planning system was decentralized, and its efficiency suffered as a result
>>1896494<I have watched kids testifying before Congress. It is clear that they are completely unaware of the seriousness of theft acts. There is obviously a cultural gap. The act of breaking into a computer system has to have the same social stigma as breaking into a neighbor's house. It should not matter that the neighbor's door is unlocked. The press must learn that misguided use of a computer is no more amazing than drunk driving of an automobile.OK Boomie.
There are various criteria for algorithms, the guarantee of meeting a criterion is easier to check for some criteria than for others. For allocation algorithms, criteria can be formulated relative to a user's own individual input in relation to what stuff is available. So you can check the veracity just based on what your inputs are and what you get for the inputs. You don't need to understand anything about programming, only what is in the pile of items to be allocated and how many other people are in the system and what weight you have. You don't need to know who of them asks for what.
>>1896665Only for simple plans. Modern supply chains are mind bogglingly complex
>>1896665>OK BoomieWhat does your quote have to do with anything?
Planning under communism is much more than just allocation, it's workers' control production. If a centralized computer system of planning is coopted it could route production in ways meant to benefit the coopters. This is computer security 101 tier stuff and you're calling me a boomer for critiquing your idea of a mainframe operated by a group of people who supposedly aren't just a new state.
>>1896889>Only for simple plans.Source: vibes.
>>1896970>What does your quote have to do with anything?Indeed. It's from the end of the PDF posted in comment
>>1896494 So I suppose you agree posting the PDF was stupid.
>This is computer security 101 tier stuffThe point of the second paragraph in post
>>1896665 is that you can check something in the results even if the program is a black box. The results are constrained by the guarantees and that means so are any possible gains for a conspiring group of programmers.
>>1900570looks cool anon, but from the website it just looks like a bunch of people who are fans of cibcom and parecon and have few speaking events
>>1900570Good concept, now let's give it traction.
>>1900811they're having a big launch event July 13th. the idea is to try and bring everyone in the planning space together
>>1900840not to be offensive, but are you sure this is not some weird cult?
>>1900851you can never be fully sure. I think the general idea is to serve as a hub for announcing seminars, new publications and such
>>1902461The "musts" are a bit strong - things like that are first to be evaluated against the planning system. But certainly Pendergrass is not the only one to suggest mandatory veganism - Andreas Malm does as well. In practice it's not possible - there will still be game that needs to be regulated, the hunting of which would result in meat. Chris Smaje (
A Small Farm Future) suggests there's plenty of room for poultry, whose walking about, poking and pooping improves the soil. Pigs also to an extent. But again these suggestions are completely detached from an actual planning system and thus wholly anarchic.
Also Pendergrass is the guy behind this:
https://play.half.earth/
>>1902461Half earth socialism is actually a pretty solid attempt at statistics based proposals. Their own simulation can be won by just going full nuclear and banning cars.
>>1902461that list is based though????
>>1902621Don't you know putting limits on unhinged consoomerism is degrowth and therefore… le bad?
>>1902630Marxism is not an austerity cult, sexually mutilated faggot. You're looking for capitalism, next door.
>>1902654That, too, is next door
>one retard arguing with another retard
>both failing to realize we might expand MoPs in one industry while scrapping those in another
you can't say either way what should be done until we have a planning system in place. arguing about this level of consumption or that is utopian
>>1902665we can have some general ideas lol
>>1902654More productivity = more free time under socialism
>>1902667you can say "we could do this", but when you start saying "we will do this" you're stepping over the line imo
>>1902483>>1902621>>1902554>>1902461Veganism is pure pseudo-science on a plethora of levels.
>>1902696pretty sure hunter gatherers mostly ate fruit and mushrooms and only had meat on occasion since hunting is hard work.
>>1902483>there will still be game that needs to be regulated, the hunting of which would result in meatthis doesnt matter, the harms veganism solves are factory farming and deforestation, not random hunters sticking within season quotas.
>>1902859sure, but then say that factory farming is to be curtailed rather than mandatory veganism which will never happen and is just ceding ground to radlibs
Unique IPs: 100