Yves here. Rob Urie looks at the war in Ukraine, the deteriorating conditions for US workers, and the seeming impossibility of generative to course-correct, in terms of widely usable, high-quality output all as compelling evidence of the US doubling down on failed policies. I’d like to qualify a bit his view that capitalism will always and ever immiserate labor. It didn’t in the US in the 1950s and 1960s when individual tax rates were high and unemployment was low, de facto giving workers bargaining leverage. But Mikhail Kalecki explained in his seminal 1943 article on the political obstacles to achieving full employment. If you have not read it yet, please be sure to do so after digesting Urie’s article. At the risk of oversimplifying Kalecki, the core of his argument is that, business owners and managers find it critical to have meaningful difference in their status and that of workers, as in higher incomes and wealth, as well as day to day power over their subordinates. Low unemployment, in their minds, fatally undercuts that.
By Rob Urie, author of Zen Economics, artist, and musician who publishes The Journal of Belligerent Pontification on Substack
Russia – Ukraine Update
Commentary regarding events in Ukraine has been limited in recent weeks by the contradictory pronouncements being made by the Americans. Donald Trump was elected in 2024 to end the American war in Ukraine. His plan, as has been revealed to date, is to mumble targeted nonsense about ‘peace through strength’ in the direction of his constituents, while imagining that his personal charm would convince Vladimir Putin to forgo Russia’s national interest and turn Russia into a resource piggy bank for Western imperialists.
From what I hear of public opinion in Russia regarding the war, Mr. Trump’s goals are decidedly not what Mr. Putin and the Russian people want. To those who imagine that Russian stubbornness is behind the current impasse, the Russians and Ukrainians had agreed to a deal that would have ended the war in April, 2022, within weeks of the launch of Russia’s SMO. It was the (Genocide Joe) Biden administration that sent Britain’s Boris Johnson to put the kibosh on that deal.
Given that the Russians and the Trump administration are still speaking (in unannounced meetings), an implied strategy is being revealed. Mr. Trump will slow-walk and prevaricate regarding continuing US support for Ukraine as the Russians finish the ground war. Once the ground war is completed, the Russians can arrange circumstances in Ukraine along lines acceptable to them and the US and the Europeans will stop mentioning the conflict and let it fade from the collective memory.
Graph: as the neoliberal ethos of ‘every person for themself’ has come to dominate the US, the percentage of alleged leaders who have served in the armed forces has collapsed. In the 1970s, over 70% of Congress had served in the military. In 2023, the last year for which data was available, this number was 18.4%. This makes the US the Chicken-Hawk capitol of the world. Source: pewresearch.org.
The weak point in this strategy is that unless the CIA and MI6 are ended, little that any American President can do will remain unmolested. Further, the current government in Britain illustrates the straight lines leading from liberalism to fascism. The Draconian pro-genocide laws passed by the Starmer administration, whereby the British government uses the full force of the state to crush opposition to its and Israel’s ethnic genocide, represents a new post-War low for Western liberalism.
In the US, the CIA has been temporarily tamped down by the threatened prosecution of former CIA Director John Brennan for leading the Russiagate Hoax. While politically motivated leaks need to be treated carefully as evidence, the rumored instances of Mr. Brennan blowing through cautionary advice to keep the hoax alive suggest that he was / is spectacularly unsuited to be a federal employee. That fans / true believers in the Russiagate Hoax viewed Brennan as a truth teller, as opposed to a cynical inside operator, is telling.
The ongoing problem for the Russians will be that the Americans are, in the parlance, agreement-incapable. This is why Mr. Trump’s solution may represent the only path forward that won’t end in nuclear annihilation. One problem with Russia claiming the territory that it has is that doing so moves the Russian border further West. This means that the only way for the Russians to create a security zone in Ukraine is to control more territory, which moves the Russian border even further West still. But this is the situation that the West has wrought, not the Russians.
Americans might have noticed a determined shift in favor of imperialist wars by the US Congress in recent years. What percentage of Congress has served in the military conflicts that Congress supports? 18.4%. This is down from 70% in the 1970s, when war wasn’t viewed in nearly as upbeat terms as the present. This makes the US the Chicken-Hawk capitol of the world. Instead of paying for Congress to take guided tours of Ukraine, why not dress it in the Ukrainian military uniforms that the US is paying for and let them spend a few weeks on the front lines?
The international political challenge that the US now faces is that it arrogantly started a war in Ukraine that it has now lost. The assertions currently being made that Ukraine can still win the war are exactly and precisely the same as those that were being made at the outset of the war. What this means is that these ideas were already tried and that they did not produce the predicted results. Why anyone would take these arguments seriously given this history suggests that no one actually is. It is just MIC propaganda.
This being America, no one is going to stand up and take responsibility for the Ukraine fiasco. Functioning nations find the culprits— Biden, Antony Blinken, Jake Sullivan, Victoria Nuland, and the heads of the Pentagon and CIA, give them fair trials, and see where the prison sentences land. However, US politicians can put resolution off for as long as the public relations wheels stay on the Ukrainian cart. Donald Trump doesn’t want to be saddled with a war that he ‘only’ armed and acted to promote. In this affair, to quote the Sex Pistols, no one is innocent.
Not formally resolving the war in Ukraine is morally and ethically cowardly because however carefully the Russians proceed, the war still leaves the world on the precipice of nuclear Armageddon. The US spent the last four months threatening to deliver dual-purpose (nuclear / conventional) Tomahawk missiles to Ukraine as the Russians are making their new nuclear-powered nuclear missile known to the world. This as the Trump administration just passed on the Russian offer to keep the New START treaty terms in effect to reduce the risk of nuclear conflict.
(Here are Ted Postol and Danny Davis discussing Russia’s new Burevestnik nuclear powered nuclear missile. Postol states his view that the Burevestnik is indeed nuclear powered— a point of contention amongst military analysts. The relevance is that it can stay in the air long enough to out-maneuver US defenses. And of particular relevance is that the Trump / Hegseth fantasy that the Golden Dome will protect the US from this and other Russian hypersonic missiles is not backed by the available evidence).
Americans need to know that according to military analyst Scott Ritter, what the Pentagon means when it claims that the US can prevail in a nuclear war is that a few of the politicians who most of us wouldn’t cross the street to urinate on if their heads were on fire will survive. Amongst the rest of us, the lucky will die quickly in the initial assault. It will be the unlucky who emerge from nuclear shelters to a world with no food production, energy supplies fully destroyed, most infrastructure destroyed, and no organized government to manage a recovery.
The partially stated rationale of the Biden administration for starting the war in Ukraine was to ‘bleed Russia,’ to box Russia in militarily as Western regime change institutions (CIA, MI6) whittle away at the Russian capacity to resist. This makes Donald Trump’s inability to advance a plan to end the war, publicly at any rate, a sign that no President can end any war without getting the Federal agencies that constitute the permanent government out of the way first. As I understand it, not even NED (National Endowment for Democracy) was actually shut down by Mr. Trump.
As I suggested a couple of months back, the Trump administration has been speaking with the Russians about ending the war in secret. Numerous stories appearing in RT (Russia Today) mention talks between the US and the Russians that weren’t mentioned at all in the Western press. In one sense this is progress. The Biden administration was incommunicative with the world’s other nuclear superpower throughout its term. In another sense, that these talks must be kept secret suggests that the US institutions opposed to peace remain undiminished.
Why Capitalist Employment Is So Miserable
One of the rarely-spoken secrets of capitalism is that it is incapable of employing all of the people who want jobs. The US narrative over my adult life, which has spanned from the late 1970s – today, is that there are always jobs for those who want them. This maneuver of claiming employment ‘abundance’ shifts the onus for unemployment onto the unemployed. It is a personal failure to be unemployed goes this logic. And through careful and persistent training, the American people have internalized this ethos.
Additionally, the myth of job abundance puts a lie to the idea that employment is determined by one’s ‘human capital.’ Unless there are jobs that match one’s ‘human capital,’ it is a wasted asset, in the parlance. For instance, in the early – mid 1970s there were rolling lay-offs of autoworkers, which was then considered a good job because it paid well and provided benefits. As the US moved to the wholesale ‘outsourcing’ model in the 1990s, millions of highly skilled manufacturing workers were cut loose to compete with 14-year-olds at French Fry stations across the country.
So, the question is, what if the broad economic frame being sold to Americans is either mistaken or a deception? From the passage of NAFTA in 1994 to today, the perception of the US economy that has been sold to Americans has been of a global economic powerhouse that represents, Voltaire’s Candide notwithstanding, the best of all possible worlds. While pretending to portray broad labor market outcomes, this economic narrative inevitably poses the lives of the richest 10% of Americans as average. But what they aren’t is representative.
Westerners are burdened with a liberal theory of employment that assumes away class struggle. In the liberal theory, ‘we’ exist in economic unity with our oligarch betters. Through their roles as good capitalists, if they can improve their lots by worsening ours, we all benefit? Yes, a scheme in which one side benefits from causing economic harm to another means that we all benefit in the liberal mind. Anyone familiar with basic arithmetic knows that this theory does not describe the social realities that we face.
Graph: in the early – mid 1990s US President Bill Clinton convinced Congressional Democrats to support NAFTA after Republican President George H.W. Bush had been unable to get pro-labor Democrats to support it. The decision caused mostly incremental changes in manufacturing employment until China was elevated within the WTO (World Trade Organization) in 2001 to flood the US with low-cost consumer goods. No functioning nation would have sacrificed its manufacturing base to satisfy the capitalist fantasies of a bailed-out Wall Street. Source: St. Louis Federal Reserve.
Oddly (not), the imperialist labor model that has informed US actions abroad for two hundred years was directed back onto the US beginning a few decades ago. Slavery was central to the American labor model for the three centuries prior to the ‘founding’ of the US. And the master / slave roles are close metaphor for modern capitalist employment. As this article from the Harvard Business Review suggests, the model for present day HR (Human Resources) practices descended from plantation strategies for managing slaves.
In contrast to the liberal model, the Marxist conception of class struggle begins with the antagonistic relationship between capitalists and employers who want to reduce wages, and workers who want to see them raised. Even capitalist icon and eighteenth century Scottish economist Adam Smith understood the power imbalance between capital and labor in his classic The Wealth of Nations. The liberal economic model (called ‘economics’ in the US) is an ex-post explanation of capitalism that was funded by the state on behalf of the rich.
I will preface the following with the view that one need not be Marxist to read a good argument. V.I. Lenin’s theory (following from Marx) of the capitalist state poses the state as the support mechanism for capitalism. Contrast this with the American Right’s (Uniparty’s) view that the state and capital are antithetical to one another. Then look at the degree of state involvement in the US in making the rich richer. Donald Trump is a rogue oligarch. And Barack Obama saved the American oligarchy when he ‘saved’ Wall Street.
Along these lines, a new report on labor market views, if not precisely conditions, tries to tie this question down. According to Gallup, the well-regarded polling organization, only 40% of employed Americans have ‘quality’ jobs, meaning that 60% of Americans are working unfulfilling jobs that don’t provide them with enough income to live. Gallup’s main criterion for the designations is income, but with ‘career’ advancement and dignity in the workplace added.
Missing from Gallup’s criteria are benefits, which typically add 30% – 40% to the cost of employment. I leave it to readers to decide if a job can be considered ‘good’ if it doesn’t have benefits. However, given that one medical mishap can end a family’s economic viability forever, the view here is that a job cannot be considered good if it does not include benefits. The likely reason why Gallup decided to exclude benefits is that it would have very quickly produced the result that far fewer than 40% of the jobs in the US are good jobs.
The political problem with going much lower than 40% is that a figure of say 5% – 10% means that very few of the bourgeois strivers looking for advancement, the constituency for liberal politics, will ever see the opportunity. People who are qualified for senior positions will never see them because not enough of those jobs exist. This is the likely explanation for why the American bourgeois are so miserable. They’ve set the trajectory of their lives up for outcomes that are unlikely to be reached.
But why don’t we stick with Gallup’s 40% number for the moment. What this means is that 60%, a significant majority of the US workforce, have crap jobs. These are jobs that do not pay, offer little dignity, and little opportunity for advancement. The way that this dichotomy was explained when I was young was that the crap jobs were stepping stones to the good jobs. The French Fry station was a step on the path to upward mobility, went the theory.
In that theory, there are ‘lifecycle’ jobs that begin at the French Fry station and end in the executive suites for talented people. I believe it was the Wall Street Journal that in the early 1990s produced a long article arguing that this is not how the US labor market works. What the Journal reporter found was that at French Fry stations, younger workers were working beside their parents and their grandparents. For 60%+ of Americans, the French Fry station is the ‘career.’
What this means is that rather than describing a unified economy which all Americans experience together, the 40% / 60% division by Gallup represents a class divide which isn’t, as a rule, traversed during the working lives of the overwhelming majority of American workers. Workers in the ‘good’ jobs were by-and-large born into the class for which good jobs were created. And workers in the not-good-jobs were by-and-large born into the not-good-jobs class.
How do we know this? The US has low and declining economic mobility (and here). What this means is that one’s choice of parents (no one chooses their parents) is the most impactful economic decision that most Americans will ever make. If you win the parent lottery, you get a ‘good’ job. If you don’t win the parent lottery, it is the midnight shift at Dunkin Donuts for you. This renders the ‘soft’ part of Gallup’s effort less meaningful than is suggested. If the class ceiling ends at the French Fry station, you can advance to running it, but nothing more.
Recent articles in the business press suggest that the minimum per annum cost of two adults, equals almost precisely Gallup’s minimum income cutoff for having a ‘good job.’ What this means is that two adults with no children will have enough income to live plus a few extras. In contrast, a single parent with one child with a ‘good’ job won’t earn enough to survive. When the household’s cost of benefits— health care, pension, etc., are deducted, net household income from a ‘good’ job yields about half of what the median family needs to get by.
Again, this would be for the 40% of US jobs that Gallup deems to be ‘good.’ Circumstances are considerably more dire for the 60% that have crap jobs. Readers are invited to discount the cost numbers if they care to account for the tendency of the business press to represent the US as being rich, when it is more accurately described as being a wasteland of poverty and people barely getting by that is ruled by a few very rich people. The business press doesn’t really report on this America.
Because Gallup’s 40% number of ‘good’ jobs doesn’t include the cost of benefits, the actual count of good jobs is considerably lower than Gallup is estimating. Further, with the debasement of the economic value of benefits via insurance companies that don’t pay legitimate claims and pension schemes designed to enrich the C-suites (executives), costs of benefits are rising as the value of these benefits is declining.
On the flip side of this, according to Forbes, fully 30% of America’s 400 richest inherited their wealth. Further, of the 70% who didn’t directly inherit their wealth, most inherited seed money, family connections, and went to the ‘right’ schools to assure that they would not be the people who fill out job applications in order to get a job. Further still, the Marxist concept of ‘class reproduction’ places the PMC as children of the earlier PMC. How do we know this? Because economic (class) mobility is very low in the US. Class reproduction is one of the products of low economic mobility.
Again deferring to the work of both Karl Marx and V.I. Lenin, the reason why economic mobility is low in the US is because of the tendency of capitalists to close the economic door behind them once ‘success’ has been achieved. The logic is that if only a few people have all of the wealth, then they are rich and we are not. If everyone has a lot of wealth, then no one is rich. This is what made Barack Obama’s bailouts of the already rich so offensive to so many Americans. Now these public welfare recipients can lord it over us by pretending that they actually ‘earned’ their bailouts.
In the late 1980s, if memory serves, a survey was taken asking Americans what it is that they feared most, economically speaking. This was during a financial boom when urban dwellers were feeling flush. Something like 90%, nine out of ten respondents, said that being homeless was their number one fear. While this likely made the Reaganite’s hearts sing— having a workforce that is broke and miserable is necessary to the good functioning of capitalism went the theory at the time, it is a startling indictment of capitalism. Who is prospering when the threat of losing it all at any moment is ever present?
The problem was, and remains, that the people who find the argument that an impoverished and desperate workforce is more productive are the industrialists, the oligarchs, and their apologists in academia who aren’t in danger of being impoverished. This is the class view from above. Most have never filled out a job application or sat through a formal job interview. They get their jobs at summer parties in the Hamptons. (I have met these people). They then have storied careers failing upwards.
Before Donald Trump ascended for a second time, the American political establishment was given the opportunity to prove that it can govern via Joe Biden. What Biden did instead was to launch a homicidal and gratuitous war against Russia in Ukraine as he launched an honest-to-God, WWII style, ethno-nationalist genocide in Palestine. This same establishment now has the empire in freefall, in economic, political, social and cultural decline.
When Americans get tired enough of this nonsense, and before the US dollar is worthless abroad, the Federal government of the US has the ability to create good jobs— well-paying jobs with benefits, performing work that needs to be done for all who want to work. The truly, deeply, uniformed pushback to this is to ask who will pay for it? Think of the math. If Federal workers produce more in value than they are paid, then a net economic benefit in capitalist terms has been produced. Many of the urban landscapes in America’s older cities were produced this way in the 1930s.
If the powers-that-be can really convince enough Americans that forty percent of US jobs are ‘good,’ what about the 60% that aren’t. What society condemns more than half of its citizens to toil for low wages in lousy conditions for their entire adult lives, and yet still considers itself successful? Answer: the one where the people explaining the place aren’t the ones toiling on the wrong side of it. Who could possibly listen to Donald Trump ‘explain’ the labor market when he has— guaranteed, never filled out a job application in his life?
A Federal Job Guarantee, as partially described above, would redistribute economic power downward, restore the simulacrum of democracy, put tens of millions of long-put-upon Americans back to work, provide the labor needed to rebuild the nation, and refocus the US away from Dr. Evil (cartoon) type strategies of world domination. This is what Donald Trump said that he would accomplish, if by other means. Well, Mr. Trump has only made the situation worse in almost every dimension there is.
Last on this point, the Democrats were apprised of this history of the US creating successful employment programs in the 1930s prior to Joe Biden taking office. What the Biden administration chose instead was to make corporate executives richer through tax incentives. Biden’s plan even called for paying the salaries of workers to build EVs, from which corporations could pay executive bonuses. The point: anyone who imagines that the Democrats are the solution to anything other than making working people poorer needs to get a grip.
What Does AI’s Low Adoption Rate by Industry Mean?
MIT has a recent paper on AI that I left on the back burner until I understood the methods better. The headline is that 95% of the corporations that have tried ‘Generative AI’ products have decided against buying and implementing them. My own take is that the language is getting in the way here. Generative AI is what I call ‘internet AI,’ the product that ChatGPT and Google AI dispense for free in the misguided belief that larger training sets will improve their products.
By calling the applications ‘industrial,’ the sense is given that the purpose of Generative AI is to improve industrial efficiency. But industry has been adding algorithms to industrial processes for the last century. I saw textile looms that could produce algorithmic patterns in woven materials in the early – mid-1970s. But this wasn’t Generative AI. As the article suggests, the main problem with Generative AI is that the training sets are unrepresentative of the processes that industries are applying it to. Recall that this is also a function of the probability-based approach that AI uses.
Otherwise, AI’s autocorrelation function could seemingly be used to correct the bias between training sets and AI output. I used to build ‘error correction’ models all of the time. If memory serves, step one is to solve for the appropriate lag. Step two is to solve for the average deviation from a local mean. Step three is to program the solution with a convergence rate. In plain language, the difference between the predicted and actual outcomes of Generative AI are used to reduce the difference going forward.
However, the ‘problem’ that doing so would create would be that the voices of power would no longer be over-represented in AI output. Mathematically speaking, this would require reducing the weight of the New York Times to less than that of The Grayzone because Times stories are republished so frequently. In fact, in addition to being favored by AI in probabilistic terms, lists of ‘authoritative’ sources are weighted heavily by Generative AI developers. In other words, Generative AI is set up to produce rigged results.
As I understand it from the MIT article, it is this ‘rigged’ nature of Generative AI that is causing the problems with industrial applications in the West. What industries need is accurate information from unbiased datasets. Input bias reduces the use-value of AI. And this information must exhibit statistical characteristics that are conducive to methods of AI. Unless US-based AI companies are producing custom training sets for different types of applications, AI will be unlikely to work as advertised.
The best guess here is that US-based AI companies are using the same unrepresentative training sets that produce Google AI and Chat GPT ‘hallucinations’ in their industrial applications. I really hope that I am correct in assuming that no one could be this stupid. The problem with doing so is that even if the relevant data is included in the AI training set, biases in both the data and the AI method will mean that statistical draws from a biased data set will hide the relevant data.
That the MIT article calls finance an ‘industry’ suggests back-office applications like automating bank account sweeps into interest paying accounts. But Generative AI seems like a wildly inappropriate tool for doing this. Even using industry-specific training sets, at best AI harvests a workable solution from the training set— called intellectual property theft within the capitalist frame. At worst, it produces the same improbable garbage that the internet AI applications produce.
That 95% of the corporations trying Generative AI products are deciding not to move forward with them suggests a large business problem for the AI ‘industry.’ Depending on who the five percent that is buying Generative AI products is, this may or may not be enough to float the massive bet that the US has placed on AI. What is relevant in the present is that this take-up rate is far, far, lower than the builders of these products were anticipating. When asked, the head of ChatGPT, Sam Altman, recently offered that the Federal government will bail out the ‘industry.’
In a certain sense it is reasonable for the MIT folk to consider the FIRE Sector (finance, insurance, and real estate) an ‘industry’—the US manufacturing base was hollowed-out decades ago. But the doubt here is that this is what the Chinese mean when referencing AI applications to industry. The information that I have been able to find (snippets) suggests that in China, industrial applications of AI are being applied to manufacturing. This ties to the history of capitalist ‘innovation’ with respect to industrial production.
In this respect, industrial applications of AI are already raising China’s industrial output. But for it to work, the Chinese likely aren’t using the same Generative AI that the Americans are selling. As the MIT article suggests, Generative AI does not work for most industrial applications. By identifying one aspect of the problem to be the training sets that Generative AI is using, this reads quite a bit like these AI companies do not know what they are doing. Question: why wouldn’t an industrial application be trained on industrial data?
Again, and with apologies for having to read between the lines here, the central problem with Generative AI is GIGO— garbage in, garbage out. MIT’s premise that simply adding more data to AI training sets will fix current problems fundamentally misunderstands the statistics at work. In fact, the biases emerge from both the training sets and the AI method. Given the evidence of this existing bias, why would adding more data reduce it? This is one problem with choosing a methodology (probability) premised in conditions that do not exist.
To state this again, yes, in theory adding unbiased data to an unbiased data set might improve AI results. But the two AIs that I have spent time with, ChatGPT and Google AI, produced biased output on every topic that I queried. As I have written, what I see looks like an authoritarian takeover of the US. Every single query result is a reiteration of state propaganda. Once you raise this point, the AI ‘fesses up and explains why its first query result will always be state propaganda— because that is what is in the training data.
