Kosambi Circle meeting 10 notes

22 May 2020 2100 – 2300 Meeting notes


This was the tenth meeting of the KRAC done online via zoom, 92 people attended. This meeting continued from the last meeting and ended our May theme of discussing technology and its interactions with workers and society from the lens of socialist theory. May had a large list of readings and in the last meeting we had covered Marx's Wage Labour and Capital, some pages from the notes of Grundrisse of Marx, (pages 690-698) on Fixed capital, Means of labour, Machine where Marx is talking about how humans and machines are used together, Kosambi's essay Science and Freedom, Capitalism’s New Clothes by Evgeny Morozov which is a critique of Soshana Zuboff's category of “surveillance capitalism”, Anupam Guha's Artificial Scapegoat, and finally we had started a discussion on Platform as a Factory by Moritz Altenried. In this meeting the first fifteen minutes people recapped and discussed all these previous readings and many questions were asked on the same, then we discussed Platform as a Factory by Moritz Altenried in earnest.

Again, some voice notes had been used to explain some of the material which can be found here: Notes on Technology

The discussion began by going into what “crowdwork” is, and why it is actual creation of value (thus differentiating it from “consumption as work ideas” which Marxists differ from). Then we went into distinctions between microwork and macrowork. The history of microwork and how it has led to crowdsourced coding etc was explored. We went into what Altenreid means by digital Taylorism (digital technology being used for the standardisation, control, and surveillance of the workforce allowing “scientific management”) and how it is altering the workforce, capitalism, and what implication that has for socialist politics. At this point we also went into what exactly “Artificial Intelligence” is and why it does not need to be “real” to be as useful political term, generally we are talking about machine learning which is sophisticated machine driven statistical analysis and that is fine. We went into some history of the Amazon Turk platform, amazon web services etc, the phrase “artificial artificial intelligence” and how as an idea it relates to the old charade of the Mechanical Turk. The idea of a “human intelligence unit” or as Bezos puts it succinctly, people as a service. Then we went into how thanks to this microwork, what Marx had predicted in terms of heterogeneity of the workforce melting away and the standard small tasks increasing the replacability and decreasing the bargaining power of the worker.

In crowdwork, several people (workers) spread over different parts of the world take part in small tasks for which the get paid. Through these tasks, data is generated. Examples were discussed of such “ghost work” and “erased work”, sometimes unpaid like captcha, mostly ill paid. We connected that work to the AI economy, such labour is then used in two ways, in the creation of data sets by codifying human intelligence like cultural nuances, to monitor the machine learning that is run, which is called human in the loop. Examples of both were discussed like automated hate speech moderation in online fora etc.

We went into something we called “the cycle of desperation”; i.e. automation taking away jobs leading to more and more workers opting to be part of the gig economic system helping create datasets and machine learning models which lead to their own obsolescence. Moritz uses the term 'deskilling' in his paper- people without jobs would keep indulging in these tasks for income- gap widens. Marx also talks about a similar system in his time piece wages, which he had reckoned would in time become the most attractive system for Capital. We discussed how thus alienation in crowdwork will increase to such an extent that impossible to determine anything about the nature of your work, and thus while it is possible for a bunch of engineers to strike in the Google Maven incident, how will strikes happen when workers don't even know their coworkers or what they are working on. This was discussed intensely. Thus at the macro scale not only is there is deskilling there is surrender of political power, labour policy, a march towards obsolescence by the working class desperate for survival, and a depoliticisation of labour. It was discussed that law, especially discrimination law is completely useless to help this because it cannot address the fundamental question of economic rights, as it takes for granted class and property. It was discussed that this steady platformisation had implications in other battles, most of the carework done by working class women, it was talked about how platforms are being made where one can hire domestic workers. This onslaught on women via a widening gender gap (and children being made to work) thanks to the expansion of piece work was something Marx had realised. It was discussed how to fight this, and a primary tactic has to be a fight for regularisation, recognition of work as work, and resistance on the dilution of labour laws. Newer forms of unionisation were discussed.

Next we discussed Lying Eyes, The Victorian roots of facial recognition by Richard Woodall, and A tale of two AI cities by Melissa Hellmann. The latter is a journalistic piece which talks about surveillance state set up in Palestine by Israel- people being subjected to constant surveillance without any evidence of any wrongdoing and how this by default aggression is considered normal. We talked about parallels in India about this de facto turnaround of the presumption of innocence in policy and how this policy creep has happened (we connected this with how technological developments also invade policy space creating a prisonification of civil life and who benefits from that), and why this securitisation has to be understood in material terms. We talked about how capital, especially finance capital dictates these developments, for one the deep relations between AI companies on the west coast and surveillance states which are treated like “laboratories” and “garages” with results like in 2018, AI companies raised 40 percent of all venture capital investments in Israel. We discussed the social credit system of China and how it has links with Chinese capitalism. We discussed how these things influence academia and what academics should do. The piece Lying Eyes is about how state surveillance started in Victorian times, how the attempt to escape from one pseudo pseudo-science of physiognomy – the idea that one's personality/nobility could be determined through one's physical appearance/bone structure/skull shape led to another pseudo-science of pathognomy by Darwin, he put forward a theory that looking at ones expression one can tell internal mental states and emotions etc and while this was an attempt to counter the essentialism of physiognomy and its racist priors but this was still unscientific and essentialist, one can do no such thing like “emotion detection” one can only watch the face make expressions and this unscientific nonsense is now given great credence in AI circles and called “emotion detection” and is considered real an authoritative – facial expressions are used to somehow 'determine' people's emotions. This led to a discussion about what machine learning actually does (pattern match from a dataset) and what if the foundational priors of creating that dataset or its annotations are false. This has severe consequences for society, in the case of “emotion detection”, this pseudoscience is used all over the world to force workers to behave in a certain way by the way of showing facial expressions, this is also used to coerce citizens, and this kind of widespread tools which makes Capital's job easier of further cowing workers leads to preposterous conclusions like in South Korea, people take lessons on how to smile, to cheat machine learning systems doing “emotion detection”. The nature of informed consent and its increasing meaninglessness when dealing with AI systems was went in and thus again the need for democratic control of certain systems and abolition of others. After this the WeWorks is a Scam piece was also revisited. And it connected nicely with the myth of geniuses solving problems, a theme Kosambi had critiqued in his piece we had discussed in the last meeting. We discussed this social distinction on how industrial capital and financial capital is perceived and how its a false distinction. We revisited the question of innovation in capitalism and how innovation will operate in socialistic states, a discussion we had in the last meeting. he point is that innovation doesn't happen in vacuum – innovation is possible only due to capital, which can and is provided by the dominant class structure – research that takes up decades is usually state-backed eg. space research.

After discussing these pieces we discussed a bit more on technology and its relation with labour and capital. Energy and energy policy was discussed and how it relates to states, imperialism etc. Past battles in the digitisation arena like Net Neutrality were taken up. The Xenofeminist Manifesto was mentioned, what it is, is it materialist or utopian, technosolutionist or not, the potential to use technology and emancipation women wrt gendered labour and care work was debated etc.

At this point the time ran out and this meeting ended. To further do these important discussions on science, technology, and socialism, the Science and Technology Caucus, was formed in the Kosambi Circle at the end of May 2020.