Home

  • Good idea: financially supporting workers displaced by AI. Bad idea: taxing companies when for displacing workers with AI.

    AI is again the topic of the day and people are discussing what to do about the coming “job-pocolypse.” It seems AI can do anything we humans can do better and so 30% or more of jobs will be destroyed and replaced by AI. Leaving aside how accurate that prediction is, if 30% of all jobs will be impacted then it does warrant a public policy response. Everyone’s got their own personal favorite, but one I see come up again and again is that companies should face a hefty tax any time they replace a worker with AI.

    To be blunt, taxing companies for replacing workers with AI is a terrible idea. Let’s leave aside the argument of “how do you prove it,” and cut straight to the fact that the government should not be taxing technological progress. Just to start with some history, how many farmers were displaced by tractors? Millions. In 1900 40% of Westerners worked on farms, now it’s less than 5%. Tractors meant that a single farmer could do the labor of tens or hundreds of men, and so they could fire many of their farm hands to be replaced by tractors. But does anyone reading this wish nearly 1/2 of us were still farmers? Should the government have heavily taxes tractors to preserve the idyllic rural farm life?

    The argument in favor of taxing companies that replace workers with a machine is that the company is becoming more profitable at the expense of the worker, and they should pay it back. The current hullabaloo is about being replaced by AI, but in the 20th century similar calls were made when factory workers were being replaced by robots. The problem with this argument is that ignores society. The worker and the company are not the only 2 pieces of the equation, society in general benefits when companies become more efficient. Technology is deflationary, and it has allows many products to drop or price or not increase as rapidly as wages in general. Food today costs less as a percent of annual income than at nearly any time in history, and a large part of that is because the cost of food is decoupled from the cost of labor. So farm hands being replaced by tractors helped all of society by giving us cheaper food, and all of society would have been harmed if taxes had been instituted to prevent tractors from becoming commonplace.

    Are the workers harmed when their jobs are replaced by AI? Yes of course. But society itself is helped and so all of society should bear the costs of helping the workers. We should of course offer unemployment benefits and job retraining to those affected. We should not let them go by the wayside the way we did to blue collar factory workers in the 20th century.

    But neither should we shoot society in the foot by blocking technological progress that will help all of us. AI replacing jobs will mean products will become cheaper relative to wages, just as what happened with food. A lot of people also spread nonsense that unemployment will skyrocket as the displaced workers can’t find other jobs. They misunderstand economics, there will always be demand for more jobs. The price of some goods will decrease thanks to AI, but that means that people can buy more of those goods or buy more of others goods that they put off buying because they were forced to choose and only had so much money. As prices fall, demand will rise, raising demand for labors in other areas, and a new equilibrium will be reached. Those jobs lost due to AI don’t mean the workers will be forever jobless, any more than 35% of the population displaced by tractors meant that unemployment skyrocketed in the 20th century. Time and time and time again technology has replaced the jobs of workers, and the workers have found new jobs. It will happen again with AI.

  • The Many Failures of Industrial Policy

    “Industrial policy” is once again the word of the day. Much like how Margaret Thatcher’s greatest triumph was Tony Blair, Donald Trump’s greatest triumph will be Joe Biden. Thatcher forced her opponents to change their policy, socialism was nixed from Blair’s Labour, and Starmer it seems unlikely to bring it back. So too did Donald Trump make free trade into a dirty word for Democrats and American politicians in general. Whereas Obama, Clinton, and 2 Bushes all championed free trade agreements, Biden increased tariffs on everyone he could (even Canada). And instead of free trade and free competition, he has made directed subsidies of domestic industry his main economic plank.

    Tariffs and subsidies make up the economic policy known as “industrial policy” and to be blunt, I hate it. Industrial Policy exactly what Servan-Schreiber proposed in his book “The American Challenge,” and I think looking at that book with a modern lens outlines industrial policy’s biggest flaws. You can’t predict the future, and so a government that tries to force its economy in certain directions often winds up funding dead ends and missing out on the next big thing. Look at what Servan-Schreiber thought governments should fund (supersonic planes, space industry), and look how much of it was bunk. Now look at all the things Servan-Schreiber didn’t think were worth a mention (genetics, the Internet, renewable energy) and look at how they’ve transformed our modern economies. And Servan-Schreiber wasn’t some rando, he was a French politician who could make actual decisions on French industrial policy.

    The government just isn’t as good as the market in actually innovating. And a hands-on government is more likely to try to smother innovation to protect jobs rather than allow creative destruction to increase productivity and national happiness.

    There’s also the inherent corruption that comes with the government funding industries. Why is Intel given so much money for making microchips when there are plenty of other chip companies out there? The excuse is that Intel is making “more important” chips, but it looks to me like they’re just plowing government subsidies right back into their dividend, handing that money to their billionaire share-holders. So billionaires receive billions of Federal Dollars, and we’re supposed to assume this isn’t corruption?

    I don’t like the government giving handouts and bailouts to their favorite, politically connected billionaires. I’d prefer companies be forced to stand up on their own 2 feet like the workers have to. You want to corner the chip industry? Do it by providing a better product, don’t just demand ever more subsidies and protectionism.

    This sort of policy is exactly the kind of failure that we learned about in Latin American history as well. Many countries in the 20th century instituted a policy of “imports subsidizing industrialization,” where they raised tariffs on foreign goods to subsidize domestic industry. This led to political capture by the industrialist however, as they realized it was far easier to protect their profits by demanding ever higher tariffs and subsidies rather than investing in producing better products. In the end these countries were left with bloated, uneconomical industrial sectors giving sub-par products to customers. The customers lost as they got less for their money than if they could just buy foreign products without the tariffs. Even today Brazil has extortionate prices on consumer electronics, higher than any other country, and the prices only go down on the rare occasions when the tariffs get cut.

    So I don’t want industrial policy, and I want it even less knowing that my political opponents can control it. America currently has a divided government, but a united government that engage in industrial policy is by far the most likely to simply hand the money to the most politically connected industrialists at the expense of everyone else. If you’re a Democrat, would you want Trump to be handing billions of taxpayer’s dollars to his favorite industries?

    And I don’t want the government to subsidize dead end industries at the expense of growing ones. I don’t want them to cut off creative destruction and leave us with hand-weavers instead of looms. I don’t want them to protect domestic manufacturers and leave consumers worse off. I don’t like industrial policy, and I think Biden’s greatest failure is that he has become Trump’s greatest triumph.

  • Quantum Computers: Hype or Hopeless?

    I’m not an expert on quantum computers by any means, but I do like to blog about things I know a little about. So bear with me.

    One of the funnest seminars I ever attended wasn’t even in my major. I’m a biologist by trade, but occasionally in grad school I’d wonder over to the physics department to eat their pizza and have a gander at the science they were touting (they occasionally came for our pizza too so fair’s fair). On one occasion, the CEO of a quantum computer (QC) company (formerly a professor of physics) came by to talk about the exciting new happenings in QC and give us some history.

    At that point I had only a surface understanding of QC, and so I had a lot of fun learning how his QC worked and the challenges he’d overcome. He also had a great sense of humor and a fun presentation. At the end he took lots of questions, and so I was also impressed by how he didn’t really sugarcoat or overhype anything. He was very open and honest that the field still had work to do and that working QCs weren’t just around the corner, whereas reading news articles you’d think they were 5 years away at most.

    When it came time for my question, I asked the only thing that I, a biologist could think of: “when will we have logical qubits?”

    To back up a big, just as classical computers store information in bits, QCs store information in qubits. Classical bits are binary, existing as either 1 or 0. Qubits are quantum mechanical in nature, and they exist as a superposition of states so that every time you measure them they have a probability of being either 1 or 0. This superposition is why QCs can do all those amazing things that people talk about like breaking encryption or what have you.

    But this superposition is also highly unstable. Any interaction with the outside world at all will destroy the superposition, rendering it useless. This has long been the bane of QC companies and researchers, how do you make a qubit that doesn’t fall apart before you can usefully use it? When the superposition falls apart, it leads to an error, and error correction in QC is all the tricks and ways that researchers are trying to either keep the superposition stable or rebuild it if it fails.

    The holy grail of error correction is the “logical qubit.” A single qubit can of course fall apart at any time so it is a poor store of information. But what if many qubits could be networked together in some way, such that if one fails the others will correct it back to its previous information value. Together, all these qubits would allow the information to be held indefinitely, even if the superposition in 1 or many of the qubits fails individually. And so together these qubits would act as a single “logical qubit,” that is a stable qubit that perfectly holds information, as opposed to the normal qubits that fall apart when you look at them funny.

    It has been theorized that a thousand or more qubits will be needed to make a logical qubit, and that the technology for networking qubits to create logical qubits is still not fully formed. So when I asked the seminar speaker how far off logical qubits were, he humbly said that they may not even be possible. From his research, quantum computers may be useful but their utility and longevity will always be undercut by the fragility of the qubit superposition.

    I was kind of stunned because in my readings on the field it was taken as writ that as soon as you can produce 1 qubit, you can scale up and produce thousands. Once you produce thousands, you now have logical qubits which will make all our QC programs work perfectly.

    What’s interesting is that IBM is now saying it will make a QC with over 1000 qubits, which is around about the number is supposed to be needed to make 1 logical qubit. Yet by in large I haven’t seen much talk about having our first crack at producing a logical qubit.

    So again I’d like to ask the question: how long until we have a logical qubit? If qubits will always be unstable superpositions, then I doubt a mass market consumer QC will ever be workable. And while the hype for logical qubits seemed ever-present when they were still a far-off dream, it seems to have subsided as they get closer to being tested for validity. I wonder if they were always nothing by hype.

  • Fitch Downgrades America’s credit rating

    I’ve spoken before about how credit rating agencies are downgrading the debt of nations. Now, Fitch has downgraded America’s credit rating from AAA to AA+. Once again I’m seeing a lot of conspiracy theories about this and I thought I’d take a moment to hit back against them.

    The first conspiracy is the common one that the financial system is conspiring against the common man. This sort of conspiracy is no different from the old “evil bankers control the world” trope, but it gets a lot more traction online when it’s framed with a leftist slant. To be blunt, the financial system is competing with itself more than it is conspiring with itself. Fitch is competing with the other credit rating agencies (Moody’s, S&P) and if it downgrades America’s credit for no reason, it would lose trust in the eyes of the financial institutions which pay for its ratings.

    Ratings agencies rate all kinds of debt, not just sovereign bonds. And financial institutions will pay for those ratings so they know where to invest and where to avoid. Trust is key to this, and without trust, Fitch would die. If financial institutions don’t trust Fitch’s ratings, they simply won’t pay for them and will take their business elsewhere. So Fitch cannot in any way downgrade ratings in a way that the broader financial market would not agree with, otherwise it would destroy trust and tank its business model.

    In this, there is a common chicken and egg problem with ratings agencies in that they usually only change their ratings when the broader market is already leaning in a certain direction. IE they are followers, not leaders. But that that just lends more credence to their ratings. The market was already very willing to believe that America needed a downgrade, so Fitch isn’t doing anything out of the ordinary. It’s the politicians who have screwed the people on this one, not the ratings agencies.

    The other, similar conspiracy I’ve seen is that the ratings agencies are conspiring to undermine Biden and tank his presidency. Biden has been trying to tout the strong economy, and some liberal commentators have been upset that the public doesn’t always buy it. So of course this must be just another GOP plot to brainwash the voters that the Biden economy isn’t awesome.

    I would point out however that American real wages are still below where they were when Biden took office. Note for example that real wages declined 3.6% from June 2021 to June 2022. That’s a big dip, and people notice it. People’s ambivalence about the economy isn’t some nefarious plot, it’s very clear when listening to people’s complaints (inflation) and looking at the data (real wages dropping). It’s also quite understandable that ratings agencies have looked at this very same data, as well as the rising debt amid partisan bickering over how to pay it. From this, they might reasonable downgrade America’s credit rating. Not everything is a conspiracy against one’s favorite politician.

    Many people will point to 2008 and the financial crisis about why we can never trust ratings agencies again. But that was over a decade and a half ago and every country added new laws to constrain financial institutions. Saying you can’t trust Fitch because of 2008 is like saying you can’t trust the Labour Party because they were in charge during 2008.

    So Fitch has downgraded America’s credit rating, and it seems the financial markets were broadly ready to believe them. Rather that stew in conspiracies, it is better to take these criticisms to heart and find a way to fix them.

  • Why are conspiracies about the stock market so common?

    The obvious answer to the question posed in the title is that the market is complicated, and therefore people who don’t understand it are more likely to fall into conspiracies than to admit their ignorance. But I truly am blown away by how common stock market conspiracies are amongst “retail” investors. I don’t just mean the meme stock traders, many many retail investors I’ve spoken to have conspiracies about how the amorphous Wall Street is driving the market one way or the other in order to “punish” someone. Usually the argument goes that X company is a threat to Wall Street, either because it supports some politics that Wall Street doesn’t or because its technology is highly disruptive to an entrenched industry. Therefore Wall Street “punishes” it by deliberately pushing down its stock price.

    Just as an aside, that disruptive tech idea is dumb merely on the face of it. The idea is usually said to me as “Wall Street has too much money invested in Y industry, and X company has a technology that could totally disrupt it. Wall Street is protecting its investment by forcing down X company.” This is dumb on the face of it: Tesla, Amazon, and Apple were all exceptionally disruptive and still grew by leaps and bounds. Just because some investors have their money tied up in Ford doesn’t mean other investors won’t give Tesla a shot. Not every investor on Wall Street is invested in the same things, so the idea that they would act as a collective unit is nonsense.

    As another aside, I wrote early about how this was a conspiracy I see a lot when talking about nuclear fusion. The idea is that fusion is such an amazing energy source that all other energy sources can’t compete, so they work together to keep it down. But if they can keep down fusion, why haven’t they kept down every other disruptive tech that upended industries through creative destruction? I never get a good answer.

    Anyway I think this widely held believe, that Wall Street “punishes” stocks and causes them to go down, is simply another case of people not understanding that the market is filled with individuals acting in their own selfish interest. The only way Wall Street could act as a collective would be if each and every investor was forced to act the same way no matter what.

    Say Company X has stock selling at 20$. Some Wall Street investors think that 20$ is a fair price for that stock. But some Wall Street investors are angry that Company X has technology which will disrupt their investments, so they want to “punish” it and force its price down to 1$. They can try to do this by selling the stock, but if the stock falls to say 15$, then the investors who think 20$ is a fair price will happily snap it up, because if 20$ is a fair price, then 15$ is a deal. Quickly the buying pressure from investors who think it’s undervalued should overwhelm those who want to push it down, and the price would stabilize.

    Now there are two reasons this mechanism could fail. One is that all investors are forced to act in concert, which again doesn’t make sense at all. Investors compete fiercely with one another, they do not work together for common benefit. And furthermore working together creates a huge prisoners’ dilemma, if even one investor breaks with the group at large, they can reap enormous rewards by buying up stock with a fair value of 20$ for just 1$. Getting to 20x your money for free is a huge incentive to break with the collective, and no greedy investor would pass such an inventive up.

    The second reason this mechanism could fail is that there are very few investors who believe the fair value is 20$, and most agree the value is 1$. But that isn’t a conspiracy in action, that’s price discovery in action. The price is an equilibrium between the expectations of the buyers and the sellers. If more and more people think its fair value is higher than at present, its price will go up. If more and more people think its fair value is lower, price will go down.

    Trying to “force down” the price of a stock below its fair value is not a profitable way of doing business. No one investor or group of investors control the market, the market is a huge competition between all investors. And so while selling a stock for a price below its fair value can momentarily drop the price, it’s also a great way to lose your own money. Meanwhile if the market is filled with investors who think the fair value is higher, they’ll buy the stock back up to the original point. All you’d succeed in doing with this trick is burning your own money.

    The price of your favorite stock went down because more investors thought it was overvalued than thought it was undervalued, not because of some huge Wall Street conspiracy.

  • The Lunar advantage

    This post is gonna be weird and long.

    I often have weird thoughts that I wish I could put into a book or story. My thought today is about comparative advantage. Comparative advantage is an economic concept that explains why people and countries can specialize into certain areas of work to become more efficient.

    For example, in Iceland the cost of electricity is very low, which is why Iceland has attracted a lot of investment in industries that require lots of electricity, such as aluminum smelting. On the other hand countries like Bangladesh have a low cost of labor, which is why labor intensive activities such as clothing manufacturing invest there. It doesn’t make sense for a company to put an aluminum smelter in places where electricity is expensive, nor does it make sense to put a clothing factory where labor is expensive. Iceland and Bangladesh have their own comparative advantages at this moment in time, and that explains their patterns of industry.

    Let’s imagine for a moment that there was a fully autonomous colony on the moon. People lived and worked there without needing to import air, water, or food from Earth. They can trade with Earth, but if Earth were cut off they could still make their own goods, just as if our country were cut off from the world we could still make our own food, drink our own water, breathe our own air. Let’s say they use super-future space technology to extract water and oxygen from moon rocks, and grow crops using moon soil.

    If there would be such a moon colony, we would assume there would be trade with Earth. Certainly the cost of moving goods from Earth to the moon and vice versa are enormous. But it was once unbelievably dangerous to cross the oceans, and people still did it because the profits were worth it. We would expect that the moon would have some comparative advantage compared to Earth and vice versa, which would make trade profitable. This comparative advantage is the same reason Iceland sells aluminum products to Bangladesh who in turn sells Iceland clothing.

    So with all this in mind, I assert that the moon’s comparative advantage would naturally be in large, heavy goods, but not because of the moon itself but because of the journey.

    Let me give another example, suppose there is a factory on Earth making steel and a factory on the Moon making steel. Let’s also say the iron and carbon for the steel can be gotten just as easily on the Moon as on Earth. I assert that the one on the Moon has a comparative advantage because of space travel. Sending goods from the Earth to the Moon means having to spend a lot of energy accelerating out of Earth’s thick atmosphere, then also spending energy to slow yourself down for a moon landing. By contrast it takes much less energy to accelerate off the atmosphere-less surface of the moon, and landing on Earth costs far less energy as you can use the atmosphere itself to brake your fall.

    So a moon steel factory can send packages of steel to the Earth at a rather low transport cost compared to vice versa. That gives an advantage to the moon steel factory, as if there are shortages on Earth the moon factory can fill them at a rather low cost, while Earth cannot do the same to fill a need on the moon. The transport costs are not symmetric, and they are in the moon’s favor. I would assert that, all else being equal, investment for steelmaking would flow into the moon and out of the Earth.

    Of course the “all else being equal” is the rub. Air, water, and food are hard to come by on the moon. Iron and carbon might be easier but all the mining equipment is already here on Earth. We would have to do a lot of work and build a lot of technology to make a moon-base even possible. But in theory economies of scale and future-technology could make it possible and even economical. And at that point it might enter a virtuous cycle due to these asymmetrical transport costs I mentioned. It will always be cheaper to send goods from the moon to the Earth, than vice versa.

    It’s just a random thought I’ve had and I want to put it in a work of fiction. In some sci-fi universe, a moon colony is economically sustained by this comparative advantage compared to Earth. But I’ve never gotten the courage to write this story so until now it’s just been an idle thought in my head.

  • Tried again, still couldn’t get into Divinity 2

    I wrote before about how I prefer real-time-with-pause to turn-based gameplay in RPGs. I wrote that post with Divinity 2: Original Sin in mind since it’s a turn-based game that I did not enjoy. I tried replaying it this week… and I still don’t enjoy it. The first island was a bit more enjoyable this time since I knew all the little places to get experience points, but honestly this game has some terrible mechanical decisions that make it an unfun slog that punishes you for exploring around the second island.

    The first decision is that characters’ power is almost entirely based on their level. Power and level always go together in all RPGs, but in Pillars of Eternity I could beat some encounters well below the suggested level just by mastery of the systems and a bit of luck. Divinity 2 doesn’t have this. A fight at 1 level below the enemy is very difficult. At 2 levels below the enemy, it’s nearly impossible. At 4 levels below the enemy I was still pulling off victories in Pillars of Eternity, and I’ve never had a chance to do that here.

    So you NEED to be at the correct level to have any success whatsoever in beating the fights. But the second terrible decision was to scatter the content for different levels all around the map with no rhyme, reason, or communication with the player. You might be wandering around completing a level 9 quest and suddenly get dropped into a level 11 fight. The quickest move here is to just reload, which also takes far too much time for a simplistic RPG like this one. There is no way of knowing if you’re heading towards content that is level appropriate or towards certain demise and another minute of looking at loading screens.

    Taken together, the game is unfun if you’re underleveled and it doesn’t do anything to make sure you’re appropriately leveled for your challenges. The game is also unfun if you’re overleveled by the way, as enemies can barely scratch you. But with nothing done to make sure I’m finding level-appropriate content, I spent most of the game not having fun because I’m either over- or under-leveled for what I’m facing.

    So yeah, they need to hire a level designer.

  • Coda to my thoughts on Civilization 6

    I recently wrote about how the AI in Civ 6 seems to be worse at its own game than previous Civ games. I want to give a shoutout to the youtuber Sulla whose “AI survivor” series put this into sharp contrast for me. Sulla sets up games of Civilization 4 that pit the AIs entirely against each other, with no player involvement. He then looks to see how it all turns out.

    In Civilization 4, the AIs will usually manage to conquer each other. The games start with 8 AIs and it’s never been the case that they all survive to reach the end of the game. On the other hand in Civ 6 I rarely see even a single AI get eliminated unless I’m doing the eliminating. It just goes to show how passive the Civ 6 AIs are, and how utterly incapable they are at using 1upt and the rules of their own game.

    Just something to think about.

  • Civilization 4 thoughts

    Playing Civ 6 made me nostalgic for Civ 4, so I made it my project to get better at it.  I’ve been watching videos from Sullla’s channel (a big Civ lp-er) and have learned a lot of good stuff that helped me in Civ 4

    Now to start with this victory wasn’t exactly easy.  I took the most broken, OP leader in Civ 4 (Huana Capac) and used a map-type that the AI does really poorly in (highlands with dense peaks, AI pathfinding screws up).  Even then I did a tiny bit of save-scumming at the beginning to fight off the barbarians.  But once things were going, I found that a Civ 4 game can be really easy… when it acts like a Civ 6 game.  Let me explain:

    The reason Highlands map is so easy is as I said the AI’s pathfinding screws up.  I never got war declared on me during my Emperor level game, my only neighbor was Charlemagne who loved me because I adopted his religion, and my other close neighbors fought inconclusive wars amongst each other with no territory changing hands.  This let me sit back and tech away, *and that’s also how I won my Civ 6 deity game*.  It’s made me definitely appreciate that one of the big things holding back Civ 6 difficulty is the AIs’ inability to conquer each other and snowball out of control.  In a normal Civ 4 game one or more AIs will declare war, conquer their neighbor, and suddenly roll right up to the player’s boarders with the world’s largest army in tow.  A successful conqueror loses relatively few units to gain a lot of territory, and unsuccessful one throws away all their production producing units.  This helps AIs snowball when they focus on conquest. I don’t know exactly why Civ 6 AI is unable to conquer each other (OK I do, it’s 1upt) but this failure is a large part of the reason why I think they aren’t able to challenge a player once you escape the early game and are in the midgame.  No single AI will be so far ahead of everyone as to be unstoppable, everyone is usually around the same size.

    So that’s a bunch of random thoughts, but it’s what I thought of when playing Civ 4.

  • Civilization (the game) thoughts

    I don’t know if I’ve blogged about Civilization 6 before.  The game has received its final DLCs and the devs have all but left to work on Civ 7, so I guess being 5 years late is the perfect time to talk about it.  Warning, this is a long post.  Also warning, I do enjoy Civ 6 and pretty much every Civ game I’ve ever played, but I will be very critical in this post.

    To step back a moment, I’ve played every Civilization game since 3.  Most Civ games have a difficulty scale with funny little names, but basically there are 8ish levels of difficulty, and the AI gets progressively more bonuses as difficulty increases.  In 3, I could barely win on difficulty 3 of 8.  In Civ 4, I could reliably win on difficulty 5 of 8, and sometimes 6 of 8 (Emperor) with the right setups.  In Civ 5, I could win on 6 of 8 reliably, and once managed 7 or 8 using a broken setup and a lot of savescumming.  

    Difficulty level 8 of 8 is always called Deity, and it is always an exceptional challenge with the AI receiving ludicrous bonuses to every single statistic.  I have never beaten any Civ game on Deity.  Until Civ 6.

    Civilization 6 was the first game I beat on Deity and the crazy thing is I don’t actually think I’m better at Civ 6 than I was at Civ 4 or 5.  I know I’m bad at Civ 3, but that’s because I hate the trading mechanics.  But with Civ 6, I genuinely think it’s just an easier game than its predecessors in an interesting and perhaps bad way.

    To start off, let’s discuss how the Civ games make higher level AIs difficult.  They don’t particularly add any new mechanics or strategies, they just give the AIs big multiples to everything they do.  At high levels, an AI city will grow 50% faster, train units 50% faster, build buildings 50% faster, and they start with free technologies.  At the highest difficulty of Deity, the AI also gets to start with 2 settlers to the player’s 1.  That means that on Deity, the AI will start with twice as many cities as the player does, and each city will be 50% more productive.  

    That’s a big hole to dig yourself out of, but the player has much better knowledge of the game mechanics and so a very good player can still win, even on Deity.  

    The thing is that the AIs understood the game mechanics in 4 and 5 a hell of a lot better than they do in 6.  Some would say 6 is more complex, but I don’t buy that, I think in many ways it has (thankfully) been made simpler and more streamlined for easier access.  But I do think the Civ 6 AI understands its own game a lot worse than 4 and 5.  

    It comes down to “one unit per tile” or 1upt as it’s known in Civ circles.  In Civ 4, you could stack as many buildings and as many units on a tile as you wanted.  Want your city to have a forge, a market, a theatre, and be garrisoned by 10 archer units?  Go ahead.  Civ 5 changed this in that only 1 archer unit can ever fit on any tile, but they didn’t update the AIs to make them good at this new system.  There’s a complex juggling act that is needed to make all your units be effective when you can’t stack them all on a tile.  And the AI is not good at this juggling act.

    In Civ 4 the AI wasn’t great but it was at least smart enough to gather a dozen units and march towards the nearest city.  If you only had a single archer in that city, well even Alexander the Great can’t win against those odds.  But in Civ 5, the AIs will gather a dozen units, and they will all get in each other’s way as they fail to march against an enemy city.  A single archer in Civ 5 can indeed pick off their enemies one by one, defeating a dozen units without taking a scratch.

    So with 1upt, wars in Civilization became heavily player-favored, as no amount of enemy numerical advantage could make up for their incompetence.  However the Civ 5 AIs still had ungodly bonuses that could let them tech up and win the game through other means.

    Civ 6 then decided to “unstack the cities,” doing to cities what 1upt had done to military units.  Now you could no longer have a forge, a market, and a theatre all in one place.  They had to be spread across the map of your city.  To make this mechanic fun, they added “adjacency bonuses” so that buildings work a lot better when they’re near things that help them.  If a market is near a river, then it can trade with far away places easier and it makes more gold.  If a theatre is near a world wonder, then it’s in a more beautiful part of town and produces more culture.  The player is encouraged to use these adjacency bonuses to get the most out of their buildings.  The AI… cannot do this to save its life.

    Just like 1upt led to the AI being terrible at war, districts led to the AI being terrible at peace.  They have no ability to manage districts or even look for the best spots to place them.  You’ll often conquer an AI city and see districts placed in just such a way that they have zero adjacency bonus, which is hard to do if you know even just look at the tooltips.  They also seem to hyperfocus on research to the expense of all else, which doesn’t really help them.

    But on deity the AI is still hard.  In fact, Deity AIs in Civ 6 are the hardest they’ve ever been, getting to start the game with three free settlers and a good sized army while the player starts with a single settler and a warrior.

    But this gives a game against Deity AIs a sort of strange difficulty curve.  On turn 1 every single AI is more than three times as strong as you because they start with 3 settlers and their cities get free bonuses.  But as the turns go on the player makes more and more good choices while the AIs make poor ones.  Eventually the player pulls ahead of the AIs, and then starts to “snowball” from there.  Snowballing in strategy games is when the strongest ones in the game get even stronger over time relative to their peers.  Players always snowball better than the AI and so once a player is stronger than the AI, they’ll never ever be weaker again.  

    So in a game against Civ 6 deity AIs, the first few turns are the hardest by far, and you can die within the first 10 turns easily.  But if you just make it to turn 50, you’re golden, untouchable even.  The AI isn’t skilled at getting any kind of victory, so even with their huge bonuses you can snowball out ahead of them and get whatever victory you want at your leisure.  This difficulty curve existed in every Civ game, but it is at its harshest in Civ 6 because the AIs have never been worse at playing their own game.

    So while I have gotten my first Deity-level victory in Civ 6, I don’t actually feel like I’m all that good at it.  I feel like I’m playing chess against a 5-year-old only they’ve replaced all their pawns with queens.  I feel like this is definitely something that needs to be improved upon in Civ 7.  “Better AI!” isn’t exactly a hype-worthy back-of-the-box quote, but these are primarily single player games and I feel the single player experience is paramount.  I think I’d enjoy my time much more if I felt that my victories were from being out-maneuvered and outplanned, rather than because my opponents got free stuff at the start.  And I think it would be more fun if I could be ahead all game and then a smart AI could sneak up and overtake me in the lategame.  As it stands, once we’re out of the classical age, I’m golden.

    I know no one at Firaxis games reads my blog, but if someone could tell John Civilization to fix his AI, that’d be great.