Ian Welsh – The horizon is not so far as we can see, but as far as we can imagine

if you built your saas business before ai, your entire business was designed around one assumption which was that software has ~zero marginal cost.

ai blows that up. every ai action costs money. incumbents now have to create more expensive tiers or introduce usage based pricing. both force customers to make a new purchasing decisions while revenue doesn’t automatically increase just cuz the product became more expensive to operate. so your cost per seat rises, your gross margins compress, & adoption remains uncertain.

meanwhile, the model labs are subsidizing ai usage like crazy, so they can undercut you. ai native startups are burning venture capital to acquire users, so they can undercut you too.

(lack of capitalization in original)

I really don’t know what to think of AI for coding. I’m no longer a coder, haven’t been for almost thirty years now, so I don’t have enough firsthand knowledge, though I did have an AI generate basic sorting and random number generation code just to get my hands a little wet. Among devs I trust, opinion is split. Some love it, some hate it, some are on the margins. Here’s one in the hate category:

Are companies actually seeing massive productivity gains from their AI adoption?…

… All of the AI projects we have observed as a team are failing. Every single one – we have seen 0% success in a year and a half, not only amongst projects we have been asked to participate in², but even within projects that we have observed in passing while doing totally unrelated work. Even if you grant that AI tooling accelerates specific workloads, the method and scale of the current investments is senseless. Frequently the failure is not related to AI itself, but rather that companies are terminally bad at running software projects effectively, and as I have remarked previously, AI projects are subject to all the failure modes of normal projects plus you can get everything right and then still fail because of the method’s novelty. Very few companies are so good at shipping software that they can afford the extra risk profile…

…

our team has quickly learned while on an engagement not to ask anything about ongoing AI projects in any context – by the time that project has started, it is too late for the management team, and intervention is not possible until a crisis point is inevitably reached. There is no conceivable positive outcome. The failure rate is so high that even basic inquiry leaves us in an untenable position. Any coherent question about how it’s going, what the goal is, who is using it, constitutes an inadvertent attack on the chain of command responsible for the work because there are no good answers to anything. Even in rare cases where my interlocutor has stated that things are going well (usually while the project is still mid-flight and failure has not had a chance to manifest), it is generally obvious that they are doomed, but at least in these cases I can simply agree and then go home to scream into a pillow for six hours straight³.

All of this is to say that I am very confident that almost every report at a company about “massive AI productivity gains” is untrue as a matter of brute fact. Even if some companies are seeing clear gains, this is the exception, not the norm.

It’s a long post and the entire thing is worth reading. This is a high end engineering consultant who gets a view on a lot of companies.

My feeling/guess is that AI is not mature, that the use cases exist, but that it introduces significant risks of new failure modes because AIs aren’t reliable and aren’t actually intelligent: they don’t know what they’re doing. If you don’t do the work yourself or audit it to the level where you might as well have done it yourself, it seems easy for errors to creep in, for hard to maintain or understand code to be created, and for new failure modes to exist.

If I were an executive in most businesses I wouldn’t be using AI for much, if anything yet, though I’d let a few people have an account and test it, and would do so myself. It’s moving fast, it clearly adds risks and mistakes that we don’t know how to mitigate yet, it’s expensive and that’s with massive subsidies, so becoming dependent on it when prices are likely to increase significantly is unwise.

There are going to be exceptions: if you’re Google, not sure you have much choice. But if you manufacture widgets or sell hamburgers or build homes or are a lawyer (who can’t afford to have the AI hallucinate cites) I’d give it a pass for now, or use it very warily in exploration mode. I use it for research myself, especially as search engines get worse and worse, but I also check the sources it uses.

And, as we’ve discussed before, non open models have huge risks, since prices can easily be raised, the government can intervene to cripple the model or deny to other nations, and the company itself can make changes to the model which make it less useful to you. So even if using AI I’d be primarily focusing on Open models (which means mostly Chinese.)

But my best guess is that this is a real tech, with real uses, which is not yet near to mature and which is being deployed before people understand what it’s good at or how to use it safely and cost effectively, or, more importantly, when NOT to use it. I’d also guess that at least so far, it’s not the second coming, the next great thing, in the way that its evangelists preach.

So far it seems to increase failure rates on real projects and raise costs significantly at the same time. Unless I’m running a business where I must be in it, I think I’d take it slow.

There are very few businesses where “the way we did it in 2022” will expose you to significant risk and costs. If you’re in one of those businesses, chill a little and observe. Let other people pay the price and make the mistakes. If you need to, because it’s the current “everyone must do” just lie and say you’re into it, while doing the minimum.

This is certainly something I could be wrong about. I’m confident that Chinese AI will win, for example, I’m less certain about how useful AI will turn out to be. I do think that I’m almost certainly right that most companies adopting it in a big way, NOW, are making a mistake. Wait. See how it turns out. Let other people figure out what it’s good for, how to manage it, how to mitigate the risks and how to drive costs down.

And remember, even if it’s the true next big thing, early rushers often get badly burned, as with the dot-com bubble. Don’t buy hype, make sure what you’re investing in is real and you understand what it’s good for and what it sucks at.

What I write here is for the benefit of everyone, but alas, I live in capitalism and I, and the site, take money to keep running. If you value the writing here and can, please subscribe or donate.

Defund The Police and Abolish Prisons?

By Ian Welsh

On August 10, 2026

22 Comments

Glenn Greenwald has an article in which he goes after AOC and others for the hysteria of the woke era (which is apparently behind us?)

Most of it I agree with, but I’m not entirely down with this:

AOC, on Sunday, went on ABC’s morning news program with Jonathan Karl, where she was asked about some of Hong’s statements, including her call for “abolishing prisons” and “[defunding] the police,” policies and phrases that have proven to be deeply unpopular. AOC, who herself aggressively advocated those policies, sought to laughingly dismiss this all as just some momentary, harmless frenzy of “craziness” that never really meant much and that everyone has all just agreed to move beyond:

I can’t really imagine America operating with no police or prisons.

But if you got rid of three quarters of prisons and released three quarters of prisoners tomorrow I think that’d be a good idea, if done with some care.

These rates are deranged, and “abolish prisons” is exaggeration for effect, but the idea is simple enough: take all that money and spend it on welfare and social services and transfer payments and so on. It’s also worth remembering that the US did not always lock up so many people:

There’s been a slight dip since 2014, but the massive spike in how many people were locked up that started around 1980 has not flattened back to the previous historical norm. American courts are so overwhelmed with cases that most people in prison never had a real trial: they just plead out under threat of massive sentences if they dared fight.

As for police effectiveness, this is a chart of what percentage of homicides they solve.

And it’s really worse than the above chart:

That statistic would have stunned detectives in the early 1960s. In 1962 the FBI reported a clearance rate of 93 percent. Today, homicide is the only violent‑crime category that clears even half its reported cases; rape, robbery, and aggravated assault fare far worse. Murders themselves declined by 14.9 percent in 2024, but the clearance rate refused to rebound alongside them.

Why the hell are more and more cops constantly being hired when the more there are the worse they are at catching criminals?

This is like everything in America. No one looks at what actually produces good results. As far as I can tell most of the decline in crime in America isn’t about cops, it’s about getting lead out of gasoline. It certainly isn’t because criminals are scared of being caught.

And this is odd. Really odd. You’d think with all the surveillance devices we now have it’d be easier to solve crimes than in the sixties, not harder, but it turns out that what really matters is whether or not the police are trusted: whether informants will talk to them. And, more and more, they aren’t trusted and most of police work isn’t Sherlock Holmesesque brilliance or cleverly uncovering forensic clues: it’s people being willing to tell the cops what happened.

So when people, especially those in crime ridden areas, start saying things like “abolish prisons and defund the police” this an indication of a problem. And part of the solution probably is to have less cops and less prisons and spend that money on other things. This is especially the case because a lot of what police do could be done in other ways: should police really be doing wellness checks given how often they wind up shooting mentally ill people?

It’d be cheaper to have social workers take over a lot of what cops do, too. Cops are very well paid in America.

I’m no fan of “woke” and I’ve been an opponent of much of identify politics for decades.

But the truth is that America’s incarceration, court and police systems are all world class piles of flaming garbage and everyone involved with them knows it.

The response to police sucking at their jobs and to America locking up more people than anyone else in the world never seems to be a common sense “this isn’t working, we should do something else” it’s always “let’s fund them more, give them more military equipment, get rid of more due process and lock up even more people.”

I’m pretty damn sure that cutting police budgets in a quarter and releasing most non-violent criminals and closing prisons, then directing all that money to social welfare and transfer payments wouldn’t make the US any worse and would probably make it better.

And if that means being “woke”, I’m to that extent woke.

Week-end Wrap – Political Economy – August 09, 2026

By Tony Wikrent

On August 9, 2026

27 Comments

Week-end Wrap – Political Economy – August 09, 2026

by Tony Wikrent

The Biggest Skygazing Day of the Century Is Almost Here [YouTube]

August 12, 2026 is the biggest skygazing day of the century. The Perseid meteor shower peaks under a completely moonless sky, the core of the Milky Way stands high in total darkness, and six planets line up across the sky before dawn.

War on voting

This Is What Military-Coup Preparation Looks Like.

WA Lawrence [via Naked Capitalism 08-05-2026]

The blueprint for a military coup begins when armed force enters civilian government and permanent command awaits a usable crisis.

A force approaching 5,000 armed Guard personnel, proposed permanent White House barriers, and a mission extending through 2029 are replacing temporary security measures with continuing federal authority.

“This is Liberation Day in D.C., and we’re gonna take our capital back. We’re taking it back.”

—Donald Trump, August 11, 2025 ….

‘Doomsday Scenario’: How Trump and Republicans Could Steal Power – Even If They Lose

Justin Glawe and Asawin Suebsaeng, Aug 03, 2026 [Zeteo]

Democrats may win enough seats to take back the House, but Republicans can challenge those wins and attempt to steal control of Congress.

The first path toward this doomsday scenario would be if Republicans replace the current House clerk. The second path is for Republicans to work in lockstep to challenge election results.

Open Thread

By Ian Welsh

On August 8, 2026

23 Comments

Use to discuss topics unrelated to recent posts.

The Continued Abandonment Of America’s Poor

By Ian Welsh

On August 6, 2026

39 Comments

Financialization and neoliberalism require everything to be looted so the rich can keep getting richer.

First:

The number of people enrolled in Medicaid and Affordable Care Act plans fell by more than 5 million in the last 12 months, according to a new report from the advocacy group Protect Our Care.

The decline stems in part from President Donald Trump’s “big, beautiful” bill — which was signed into law last July — and the expiration of the enhanced ACA subsidies, the group says. The law includes nearly $1 trillion in Medicaid cuts over 10 years; the subsidies expired in December after Republicans declined to extend them, leading to double- to triple-digit premium increases for millions of people.

This general trend is exactly what I predicted during the debate about the ACA: it would initially help, it would become more expensive over time and it would slowly die from a thousand cuts.

Second:

Using data from the U.S. Department of Agriculture, CBPP, a left-leaning policy organization, found the number of people on SNAP dropped by 6% — or 2.5 million people — from July, when the bill was signed, to December. In Wisconsin, participation fell by 2.3%, which equates to more than 15,500 people.

The post war consensus was to take care of the poor and people who the economy didn’t work for. Every since the 70s, and especially Clinton (with his welfare “reform”) every part of that architecture has been chiseled away. At one time, by law, all medical institutions were not for profit, including health care. That was repealed in the 70s and the predictable result happened.

Not for profit or mutual insurers are much cheaper. One of the few formal acreditations I have is “Fellow of the Life Management Institute”. Despite the name, it includes a lot about annuities and health insurance, and plenty of cites for studies. (Or did back when I took it.) And the studies are clear: for profit insurance is more expensive. This shouldn’t be a surprise, after all, they want to optimize profits, not care.

Cuts to SNAP have been going on for decades, and except during the pandemic all I remember is cuts and restrictions.

If you’re poor, the government will keep offering less and less help. If you require government assistance to live, or live half decently, you’d better find another way, I’m afraid.

This is also true Medicare and Social Security, which have been nibbled to death by constant under-statement of inflation.

Meanwhile the military budget keeps going up, the Pentagon can’t pass an audit, the US is running out of weapons, Ukraine is out of Patriots and being hammered into the ground, Iran is winning against the US and heck, the US couldn’t even with vs. Yemen’s Ansar-Allah.

If America doesn’t have health care because it has a big military, Americans sure aren’t getting their money’s worth.

But because of the Red Queen’s Race that rich are in, where if they don’t get richer at least as fast as other rich, they get bought out and drop out of the game, nothing is off the table when it comes to cuts. After all, those missiles and weapons have MASSIVE markups and high profit margins. SNAP? Helps some farmers, but otherwise no one’s making money off it, and the margins are low. Medicaid, same idea: low margins.

Surely there are wealthy people, or governments, who can earn the rich higher profits than poor people, so fuck them.

This is the West. This is the legacy of neo-liberalism. All because American whites didn’t want to be forced to live in the same neighbourhoods as blacks, so they voted Reagan. (That was the margin of victory, check into it yourself.)

If you’re in trouble in America, don’t cross the kill line, because once you do, no one’s going to help you enough to matter.

Understand China’s Plan To Become A Full Socialist State

By Ian Welsh

On August 5, 2026

28 Comments

“The banks, as we know, are centres of modern economic life, the principal nerve centres of the whole capitalist economic system.”

–Vladmir Lenin

Deng had four principles for his changes to Chinese Communism:

Upholding the socialist road (or socialist path).
Upholding the people’s democratic dictatorship (originally often referred to as the dictatorship of the proletariat; the phrasing was later standardized as “people’s democratic dictatorship”).
Upholding the leadership of the Communist Party of China (CCP).
Upholding Marxism-Leninism and Mao Zedong Thought.
Chinese and American flags flying together

In Marxist thought the capitalist industrial stage is necessary before the final condition of a state completely controlled by the workers, without class.

This is because a significant surplus is required for everyone to have a good life, and it’s capitalist industrialism which creates that.

You can easily create a heavy industrial state in a command economy: both the Soviets and the Chinese under Mao did so. What they could not do was create a state with a lot of consumer goods.

Mao created the pre-conditions: heavy industry, basic infrastructure, and an educated people.

But notice the principles are all “stay socialist and stay in charge.”

The major banks in China are state controlled. The banks that aren’t state controlled are under strong supervision, and in most cases have party members in leadership positions.

Banks (and various shadow banks in the West) are the main way firms get money. If the banks won’t loan, your business won’t happen or won’t survive. If they will loan to you at preferential rates because you are doing what the Party wants, usually based on the five year plan, then you’ll have a good chance of success.

When I wrote my article on principles for running a left wing government I included taking control of the banks and taking control of media as absolute, immediate, necessities.

China has kept control of both.

This issue became especially urgent after the fall of the USSR and the CPC spent a LOT of time studying why the Soviet Union fell.

But despite Deng’s principles what was happening before Xi was the corruption of the CPC and its mingling with the new capitalist rich: after all, if as a party official you helped a firm, they could kick back to you and you’d get rich. Do enough of this and you could become obscenely rich.

This imperiled all four of Deng’s principles, as decision making around loans and regulation was increasingly made not in alignment with the CPC’s goals, but to make people rich. (One example is the massive housing bubble which Xi has since popped.) This is not the issue the Soviets had specifically, but is related: people with power no longer really believed in Communism or put the Party’s power first. (Gorbachev deliberately removed power from the party.)

When Xi took charge his first and most important programs was anti-corruption. At the time I cynically assumed it was a way of taking out his enemies, and that he would tolerate corruption for his faction.

I was, mostly, wrong. Xi’s anti-corruption is real. He’s a sincere Communist who wants to keep the CPC genuinely in charge and not let it be corrupted.

The goal is simple enough: stay in charge, make China vastly prosperous and transition to full socialism. Markets are useful for this, but must be kept under control or they will take control of the party, and capitalists aren’t going to be committed socialists.

Xi, in this light, is just an extension of Deng: actually making Deng work. Deng’s policies caused rapid technological advance and improved light industry massively, but the principles were not actually being enforced.

Xi enforces them. Not just anti-corruption but changes in how the party recruits and trains and promotes, all intended to keep the CPC on the path to socialism.

Will they succeed? Well the transition will come, if it does, with the next leader, most likely. Xi is putting his successor in a good position. AI and robotics are being directed towards physical activities, there are dark factories with no workers, and so on.

Genuine socialism means people shouldn’t have to work a lot if they don’t want to, and should still live a good life. AI and automation, done properly, offers that possibility.

Of course huge problems around distribution and production remain and will have to be figured out. If China stays the course they’re trying to create something which has never existed: something genuinely new. Westerner’s might call it “fully automated luxury communism” though I don’t know if it’ll get that far.

But the ducks are in a row: massive productive capacity, ways to reduce the amount of workers needed, and keeping control of politics and the economy.

This is the plan. No one knows if it will succeed, but it’s not a stupid plan or one that will obviously fail, IF China can deal with resource depletion, ecological collapse and global warming (all of which it is doing more to mitigate than any other country, though it’s far from perfect.)

China is the future. It’s success or failure will determine what political/economic models are considered usable and successful. The age of liberal democracy is near its end.

America Almost Out Of Long Range Strike Munitions

By Ian Welsh

On August 4, 2026

21 Comments

Why did Trump call off his big attacks? Why did the US stop attacking suddenly?

he U.S. Army has used up much of its stockpile of highly accurate long-range missiles during its five-month war with Iran, according to three people familiar with the data, raising concerns about the military’s readiness for future conflicts.

The missiles are principally the Army’s surface-to-surface weapons, known as Army Tactical Missile Systems (ATACMS) and Precision Strike Missiles (PrSM). The U.S. has used “virtually all” of these weapons, according to two of the sources.

There are also claims that the US is running out of long range air to ground munitions as well.

Combined these two things mean that if America wants to keep bombing Iran, they have to actually fly over Iran and put themselves in much more danger from Iranian anti-air, and by all accounts the Iranians still have plenty left.

This problem comes from the same place virtually all US problems come from: rather than optimizing for results (production) everyone optimizes for profits. Selling fewer missiles for much more money is a better business deal, yes?

And that the US cannot significantly increase the production is due to de-industrializing: sending most of America’s key industries to China, because hey, it doesn’t matter where things are made.

China is now the US in WWII. They could, in a war, easily retool to produce literally millions of drones. They can produce a hundred warships per warship the US produces. Etc.

These are systematic issues. De-financialization is politically impossible. Without de-financialization re-industrializing is not possible, as it would require companies to make far lower profits. Chinese companies are MUCH bigger than American ones, and make less profit (not revenue, but profit) and have far lower valuations.

In 2025 BYD sold 2.26 million cars. Tesla sold 1.64 million. And China has many auto companies.

The mindset for scaling and the mindset for profit-maxing are very different. American elites are all looking for moats: they all want to be able to charge more for a product than it is worth because no one is competing with them on price. (This is why Chinese EVs aren’t allowed in America: they compete on price and the American car industry would cease to exist on contact.)

America lost the Iran war over fifty years as it changed from an industrial power to a financial power. Power comes from what a country can actually do. What it can buy from other countries is a minor consideration unless it has the ability to force them to sell, and that ability comes from what the country can do. Against Venezuela and Cuba, the US is still strong enough. Against a middle power which prepared for war and knows how to use geography, not so much.

In general terms, everything that is happening was predicted decades ago by objective people looking at the trends. Power comes from capability, America was systematically getting rid of that capability to optimize for the number of rich people it could generate.

Military decline, industrial decline, technological decline: all of this comes from making profits the end-goal, rather than the mean by which worthwhile goals were achieved.

“We need to make a profit do X” is very different from “we do X to make a profit.”

Facts On the Ground In Ukraine

By Ian Welsh

On August 3, 2026

40 Comments

Escalating against a more powerful foe is usually a bad idea.

Ukrainians are finding this out now. There’s no denying that Ukraine’s strikes on Russia oil infrastructure, especially refineries have been effective, or that its strikes against Russian shipping had an effect but that campaign led to Russia changing its strategy as well.

It has basically turned Ukraine into a landlocked country with massive strikes against Ukrainian ports. This is as the harvest needs to be sold.

A systematic attack on gas and diesel distribution in the country, by taking out gas stations, has made logistics within the country extremely difficult. It’s not just that Ukraine is landlocked, it’s that once you get into the country by land, it’s hard to move at all.

This has always been the problem Ukraine faces: Russia is just more powerful. Putin was very restrained for most of the war, not taking out dual use infrastructure, letting power stay on, etc, etc… This is the sort of stuff that the US usually takes out first, as it did in both Iraq wars. Russia didn’t.

But while Putin has been remarkably humane for a wartime leader, there is an implicit quid-pro-quo here: we leave certain things alone and you return the favor. Ukraine, knowing it was slowly losing by accepting that status quo has broken it multiple times and each time Putin’s release of constraints on the Russian military has meant that Ukraine was hurt worse than Russia.

It’s instructive to notice that in the Iranian war the US has actually been fairly restrained compared to its normal operations, for the simple reason that while Iran can’t hit the US homeland, it can hit US allies infrastructure. So Trump’s latest promise to turn off all the lights in Tehran came to nought, as have the past threats, because Iran can turn off the lights and water in the Middle East, and more importantly, smash oil production and refining in ways that would take a decade to repair, even as US petroleum reserves keep dropping.

John Arnold put together a good timeline:

March 23: Postponed announced attacks on Iran’s power grid, citing “very good and productive” negotiations March 26: Extended deadline by another 10 days; negotiations were “going very well.”

April 7: Two week ceasefire to give time to conclude a proposed 10-point plan April 21: Extended ceasefire indefinitely after mediators requested more time May 18: Held off a major strike because “serious negotiations” were underway

June 11: Canceled strikes hours after threatening to hit Iran “very hard,” claiming a negotiating breakthrough

June 17: Signed an interim ceasefire agreement ending hostilities and beginning a 60-day period to negotiate a final agreement

July 27: Paused bombing campaign to give diplomacy another chance

August 1: Canceled planned attack after claiming that Iran had agreed on parameters of a deal

Iran says there’s not only no deal, but they aren’t even talking.

So deterrence has, to some extent, worked in the Middle East, but it hasn’t worked in Ukraine.

Zelensky is now offering Russia an “air campaign” cease fire: they’ll stop hitting Russia’s oil infrastructure and ports in exchange for the same.

Too late is my guess. “Now that this strategy is hurting us more than you, let’s stop” isn’t likely to be all that convincing to Russia.

Ukrainians are resisting conscription massively, the Russian army’s advance is slow, but it continues to advance, and Ukraine’s almost out of interceptors, since they’re going to the Middle East and Israel. A promise to let Ukraine make its own patriot missiles is unlikely to work: if they build them in Ukraine, they can be hit. This isn’t like drone manufacture, you can’t farm it out to thousands of dispersed households: it’s need a real factory, and the Russians will find that factory and hit it. Oh, and Ukraine’s post-war population is likely to be ten million less than the pre-war population, with women massively outnumbering men.

Ukraine needs a comprehensive peace deal. Much like America in the Middle East, this is a war it cannot win and which hurts it more than it hurts Russia. When you can’t win a war, you make concessions and go to peace. Denial of reality is not a war plan, it’s a way of marching to doom. At some point the Ukrainian army will collapse, and the conditions that Russia imposes in an unconditional surrender scenario are a lot worse than those they’ll get, even now, in a negotiated peace.

The same is true in the Middle East. Yes, Iran can’t conquer America, but what happens when diesel and gasoline and fertilizer and plastic and medicine shortages all hit the US mainland? The deal America gets in that situation is worse than the deal it gets if it makes peace now.

When you’ve lost a war, you’ve lost the war. Making your opponent prove it across the board will never lead to a better outcome.