Tobold's Blog

Thursday, June 26, 2025

The illusion of thinking

This month, Apple published a scientific paper about Artificial Intelligence titled “The Illusion of Thinking”. While some people pointed out that Apple wasn’t doing so good with AI, and concluded that the paper was “sour grapes”, with me the paper resonated, as it describes what I believe: That AI isn’t actually thinking, but rather just matching patterns. Which makes AI good at solving problems that have frequently been solved before, but bad at original thinking. For example the researchers showed that AI could solve the Tower of Hanoi puzzle only to a certain number of discs, while a thinking human who found the solution for a small number of discs can extrapolate the solution to all numbers of discs, as the solution is iterative. This coincides with other research which shows that the new “reasoning” LLM models are even more prone to hallucinations, that is making stuff up when they don’t know the real answer.

While Apple might have some self-interest in showing that AI isn’t that easy, the companies that push for AI have a much bigger interest in keeping the hype up, up to levels that might be described as a scam. Companies like Nvidia have their share-price directly depending on the belief that you just need to scale up LLM models far enough to reach artificial general intelligence. I consider it likely, that this premise isn’t true, which might at one point burst the AI bubble and destroy billions in investment.

At the same time, AI is growing closer to becoming smarter than humans by the clever trick of making humans less smart. The first studies of the effect of the introduction of ChatGPT on education are devastating: Students using AI to write their homework are less smart in performing similar tasks themselves to the point where a diminished number of connections was evident in their brain scans. As a large percentage of students use AI, we will need to fundamentally rethink our education systems, and fast. “Homework”, where teachers rely on their students to let’s say write an essay on their own, might become a thing of the past, replaced by more educational tools like supervised exams, where the students can’t use AI.

- posted by Tobold @ 8:24 AM Permanent Link (2) comments

Monday, June 23, 2025

Clank! Legacy 2: Acquisitions Incorporated – Darkest Magic

Yesterday me and my wife started a campaign of Clank! Legacy 2: Acquisitions Incorporated – Darkest Magic with another couple we regularly play board games with. In fact, we had played the first Clank! Legacy game with them. We all liked it a lot, which is why I then backed the crowdfunding campaign for the second game. Starting the new game was fun again, and we are looking forward to playing through the whole campaign.

Having said that, there is an obvious problem with the Clank! Legacy games, which is in the title: They are true "legacy" board games, which means that you can't play through the campaign without massively modifying the game in the process. Which then means that you can't play the campaign again, unless you buy a second box of it. It's a hundred bucks plus shipping, for 12 sessions, and then the box becomes basically useless. Technically you can play the game in its end state, but it is then static, and not much fun anymore. There is no way to reset it, and that also makes it impossible to sell the game second hand.

In fact, I am wondering what to do with the two boxes once we finished the second campaign. I kept the first box after the first campaign, but two boxes are taking up noticeable shelf space. Still, I would feel bad about chucking the boxes into the garbage bin. Probably it'll end up on the attic, with all the other stuff I never use, but can't bring myself to throw away.

Labels: Board Games

- posted by Tobold @ 1:08 PM Permanent Link (0) comments

Friday, June 20, 2025

A downgrade regarding Switch user profiles

I had 4 different user profiles on my Nintendo Switch 1: My main profile, a second profile that allowed me to play the Legend of Zelda games with a different save game, and two profiles for members of my family. When I installed the Switch 2, the 4 user profiles got transferred too. So I was thinking of using the Legend of Zelda save game on my second profile for the item transfer function of Zelda Notes. Turns out, you can't. Or rather, only one user profile on a given Switch 2 console can be linked to a given Nintendo account. All other profiles are strictly offline only, and if you try any functionality that requires online, you'll be asked to link to a Nintendo account, and end up with error code 2124-0150 informing you that a Nintendo Account cannot be linked to more than one user on a single console.

While the Legend of Zelda games previously had only one save game slot per user, I could in fact have as many save games as I wanted, by creating as many user profiles as I wanted. The Switch 2 upgraded Legend of Zelda games have 2 save slots now, but only one user profile can use Zelda Notes or other online features like game chat. If I want more than 2 save slots and create a user profile for that, I can only play offline, and without the very useful Zelda Notes added functionalities. Bummer!

- posted by Tobold @ 2:30 PM Permanent Link (1) comments

Thursday, June 19, 2025

Link's weird progression

Since I got the Switch 2, I have mostly been playing Legend of Zelda: Tears of the Kingdom again. As I had already played through the main story twice when the game was released two years ago, this time I was more concentrating on everything else. And that led me to realize how different the progression system in the Legend of Zelda: Breath of the Wild and Tears of the Kingdom is to other role-playing games.

In very many analog and digital role-playing games, your character progresses by gathering experience points, which give levels, and gaining levels makes you stronger. Then a higher level character can tackle higher level enemies. Depending on the game, the higher level enemies either just appear as you level up, or you have different zones of different levels, and gaining levels gives you access to these higher level zones. The Legend of Zelda games work completely different: If you kill a lot of monsters, especially bigger ones and boss monsters, a secret, hidden xp score goes up. That xp score levels up the monsters all over the world. But *not* your hero. Meanwhile many of the activities that would make your hero stronger, like doing shrines or gathering materials to make better weapons, are not directly linked to monster fights and the xp score.

As I don't enjoy action combat all that much, I am avoiding combat as much as possible. So the first thing I did in this Tears of the Kingdom run was to unlock all the sky towers. That not only reveals all of the surface level map, but also gives a convenient network of fast travel points. Use the sky towers to launch up into the sky, and you can both reach various sky islands, and fly very far on the surface level to reach shrines and other points you want to go. With very little walking comes very little combat. Which means that my hero is now rather strong, due to having gathered a lot of stamina and hearts from shrines, and having gathered good armor and weapons. But my hidden xp score is probably still extremely low, and the monsters haven't evolved much. So when I do get into a fight, it isn't much of a problem.

The only flaw in this strategy is that a bunch of features are locked behind doing at least one of the four regional boss fights. So sooner or later I'll have to go to Rito Village and do a part of the main quest there. Other than that I have effectively decoupled my progress from the monster scaling mechanic.

- posted by Tobold @ 9:12 AM Permanent Link (1) comments

Tuesday, June 17, 2025

I might be that guy now

In the board game community where I am frequently playing, the most played game is probably Dune: Imperium. I have played it a lot. And more often than not I played against that one guy who is really good at Dune, and nearly always wins. With time I got better, but from all the games I played against him, I only won twice. Dune is a game which ends when somebody reaches 10 points, but the winner is the one who has the most points at the end of that round. That guy usually managed explosive last turns, making 5+ points in that last turn and winning with 14+ points, when the second player had maybe 10 or 11.

Now that guy moved away, for a job opportunity in another city. And this board game evening, I was invited by some other people who wanted to play Dune: Imperium. And so I played with them. Playing against that guy had taught me a lot of things. So I built up my resources, and managed to make 6 points in the final turn, beating the second player with 15 to 10 points. And then I realized that *I* might have become "that guy" now, having inherited the title by him moving away.

I'm not sure I want to be that guy that always wins. I probably play less Dune in the future. Playing to win is fun, but winning because the better players aren't around anymore not so much.

- posted by Tobold @ 10:41 PM Permanent Link (3) comments

Thursday, June 12, 2025

The price is right

The Switch 2 is now the fastest selling console of all times. And for me, there is a lesson in here somewhere about basic economics. A lot of people said that the Switch 2 was "too expensive", but I actually think that the price is right. How do I know? There are fewer scalpers around, and they aren't having the field day that the Playstation 5 gave them.

Lamborghini makes about 10,000 cars a year, and sells them all. If they sold their cars for half the price, they would still sell the same number of them, and just make less money. Lower prices only make sense if there is an oversupply (or infinite supply, like a video game), and selling more at a lower price makes more money than selling less at a higher price.

The PS5 launch was bad economics. Demand far outstripped supply, leading to scalpers shifting thousands of units. The problem is that this doesn't make the console cheaper, because customers still pay a higher price; and the money isn't going to the maker of the console, who could invest that money into making better hardware and software, but to scalpers, whose earning will never benefit the gamers.

Since the end of the pandemic-induced peak demand, gaming news have been full of stories of layoffs and studios closing down. Which is to say that currently the industry is often spending more money on making games and hardware than what they are earning. There aren't any good solutions for that. Either companies need to reduce costs, e.g. by using AI instead of humans to make games, or simply making cheaper and less good games; or they need to earn more money, e.g. by raising prices. Neither of these solutions will be popular. But the popular solution that we have seen so often in the tech sector is that the customer is getting more than he paid for, because investors sunk so much money into a company that is operating at a loss. That isn't a sustainable business model either.

Somebody saying that the Switch 2 is "too expensive" just means that it is too expensive *for that person*. An economic "too expensive" would be if the consoles would sit unsold on the store shelves, which clearly they don't. Economics in general is about how to distribute limited supplies, and while capitalism sure has its flaws, it is up to now the best answer humanity has found for that question. The people who want the console the most are identified by them being willing to pay for it. That would work better in a system with less inequality, but that isn't exactly Nintendo's fault. Me, I gladly gave my money to Nintendo for that Switch 2, and hope they are using the money to make more great games for that console.

- posted by Tobold @ 8:06 AM Permanent Link (5) comments

Wednesday, June 11, 2025

That felt outdated

Yesterday evening, at my board game night, I ended up playing Betrayal at House on the Hill (3rd edition) twice. I don't think I got the full experience, because through the random selection of what actually haunts the house on the hill we ended up with both scenarios having no traitor, and just being cooperative against a common threat. And then the second game ended early for me, when the haunt on its first turn one-shotted my character, and I was out of the game at the earliest possible occasion the rules allow.

While it was interesting to play a classic board game, which was originally designed over 20 years ago, much of the design felt outdated. It is a game that very much follows the "Ameritrash" board game design philosophy, which is very thematic, but also very random, and doesn't allow for all that much long-term strategy. These days, even American board game designers do at least a hybrid, and allow for more player agency.

I think it was the lack of player agency more than anything else that made Betrayal at House on the Hill feel so outdated. The choices given to the player felt meaningless; for example you get to choose through which door to go, but then draw a random tile to determine the room behind the door, and quite possibly a random card indicated on that tile. I wasn't so much playing the game as rather experiencing it; stuff happened to me, up to my death happening to me, without giving me much control over what happened or very much I could have done differently. I died because the haunt rolled high on an attack, I rolled low on defence, and the difference was applied as mental damage, being more than what I had in mental stats, in spite of still having my starting stats. The same attack could have ended not damaging me at all, if I had rolled high, and the haunt had rolled low. It was just pure random bad luck.

As this was already near the end of the evening anyway, I was just able to leave. Which was better than still waiting for at least half an hour and watch the others finish the game. Getting eliminated early is a game design element you don't see very much anymore. But the high randomness and low player agency bothered me a lot more than losing the game did. The next time somebody invites me to a game of Betrayal at the House on the Hill, I will politely decline. Having said that, this might be more of a problem for people like me, who play a lot of board games. The lack of planning might actually be a feature for people who either play a lot less, or don't like to think hard when playing.

I have a feeling that today's board game night is going to be a complete opposite experience. I am going to play Aeterna for the first time, and for all I hear this is a game that is very much about advance planning, with very little randomness. There is a random distribution of cards at the start, but then those cards are drafted, which diminishes the effect of randomness; and for the rest of the round there are no other random elements other than the unpredictability of your fellow players. What makes Aeterna special, is an Unrest mechanic, which punishes overextension. Play too risky, and you could end up turning your victory point lead into a total loss due to maxing out Unrest; play too conservatively, and your Unrest will be fine, but you won't make all that many victory points.

Labels: Board Games

- posted by Tobold @ 10:14 AM Permanent Link (2) comments

Saturday, June 07, 2025

Zelda Notes is game-changing

Both Legend of Zelda: Breath of the Wild and Tears of the Kingdom have a $10 upgrade for the Nintendo Switch 2. You get them for free if you are subscribed to Nintendo Online, but I never was. So I paid for the upgrades, and was mostly expecting prettier graphics, higher frame rates, and faster load times, all of which I got. But I also got Zelda Notes, which is part of the Nintendo Switch App on iOS and Android. And it turns out, that this actually changes the games quite a bit.

Besides a bunch of smaller functions, Zelda Notes has two major additions to both games: A navigation app, and an item sharing function. The navigation app is an interesting compromise between the original version, in which you had to discover everything by yourself, and an "Ubisoft" type of open world, where every point of interest is marked on your map. The navigation running on a separate device makes it less intrusive, and you choose yourself what types of points of interest you want to have on that map. For example, I am mostly just using the map for the Koroks. In Tears of the Kingdom there are 800 Korok locations, and many of them are not obvious to find. For example there are a lot of Koroks hidden under rocks, but there are even more rocks in the game without a Korok under them. As you need the Korok seeds to enlarge your inventory, previously you had to either literally leave no stone unturned, or use a third party map app. Having this now as official part of the game is great, especially since the app can also remove the found locations from the map automatically, being linked to the game.

The item sharing function didn't sound all that interesting, until I tried it. I don't have anybody to share game items with. But the name is deceiving, the function can be used for a lot of other things. For example I started a new game of Tears of the Kingdom, and used the item sharing function to transfer some items from my old save game to the new game. With some materials being hard to farm, I really didn't want to do all that farming a second time, and item sharing provided a convenient workaround. But you can also share items just with yourself, which transforms the function into an added inventory, which is extremely useful in the Zelda games. Everybody who played Breath of the Wild and/or Tears of the Kingdom had those situations where he found another weapon he liked, but his weapon slots were all already taken. Now you just send some lesser used weapons from the game to your mobile device, until you need them and send them back. I haven't tried it, but I was wondering whether you couldn't actually use that to duplicate items, by making a save, then sending an item, then reloading your save with the item now both in your inventory and on the mobile device.

The rest of the Zelda Notes app is cute, but with less impact. There are some added voice memories, an achievement system, game stats that can be compared with global play data, a daily "wheel of fortune" game bonus, added support for amiibos, and a photo studio. There is also an autobuild sharing function, which has potential, if Tears of the Kingdom still has enough active players to make this take off. Building complicated machines is a bit fiddly sometimes, and being able to download other player's blueprints and autobuild them could be quite interesting.

All in all, I didn't regret paying for those upgrade packs. And as I said, all together it encouraged me to start another run of Tears of the Kingdom. I have played the game to the end in 2023, but hadn't explored all parts of the map, and certainly not all Koroks. Maybe a map app was what I needed.

- posted by Tobold @ 10:34 PM Permanent Link (0) comments

Thursday, June 05, 2025

A console premiere

I am not the world's biggest console fan. As a result, I am not usually rushing out to buy the latest and greatest new generation. I never got a PS5, and never owned any version of an XBox. The Nintendo Switch, released in March 2017, I got for Christmas of that year. So today is a premiere: I just received and am currently installing a Switch 2, on release day. The very first time I get a console on the day of its release.

Installation up to now went relatively well. One detail I missed was that my old Micro SD card from the Switch isn't compatible with the Switch 2, so I needed to order a Micro SD Express card from Amazon for delivery tomorrow. But there is a way to transfer my system from the old Switch to the new wirelessly, although it was a bit fiddly to set up, and required me to first update both consoles.

Did I absolutely need a Switch 2? Probably not. But it is my preferred mobile console to play on holidays. The backward compatibility with my old library of games was a huge selling point for me. But I have still to look into buying "upgrade packs" for older games, that make them prettier on the new console. Ultimately, I bought the Switch 2 because I believe that future games I want to play, like the next Zelda, will come out on the new generation of the console.

Availability of the Switch 2 here in Belgium was good. I preordered the console from a Belgian electronics store with no hassle at all for €470, and got it delivered to my doorstep on release day. No need to stand in line anywhere. I can see the Switch 2 still being listed as available on the same website, as well as on other European electronics online stores. I don't know how the availability of the Switch 2 is in the US, due to the time zone difference it is too early to get news. But I am wondering how much tariffs Nintendo had to pay for the consoles made in China, and how much that affects the availability in the US. If their profit margin is much lower or negative at $450 due to tariffs, and supply is limited, would they ship that supply preferably to Europe?

- posted by Tobold @ 10:43 AM Permanent Link (4) comments

Wednesday, June 04, 2025

AI doom

It is a sad reality of today's content creation economy that the more outrageous and disturbing your claims are, the more clicks and therefore revenue you generate. One of the recent widespread subjects is based on a prediction by Goldman Sachs that 300 million jobs could be replaced by AI. Thus various content creators are making AI doom videos, saying how we all will be unemployed in the future. Nobody mentions that we have been there many times before, and it turned out that this is not how economy works.

Today's AI doom predictions of widespread unemployment very much mirror what was said by the Luddites between 1811 and 1816, railing against spinning and weaving machines, and often breaking those machines in protest. But rather than leading to economic doom, the machines ultimately grew the English textile industry by a factor of three over the next 50 years. Jobs changed, but people moved to better-paying and less back-breaking jobs in the industry.

"Bullshit Jobs: A Theory" is a book by anthropologist David Graeber, published in 2018. The book explores the phenomenon of work that is perceived as meaningless, unnecessary, or even destructive, despite generating income for those involved. Graeber argues that a significant portion of modern jobs fall into this category, and these jobs can negatively impact individuals' mental health and social well-being. And guess what, these are exactly the jobs that are first in line for being replaced by AI. AI is best at repetitive tasks that do not need much human judgement, expertise or decision-making. The bureaucratic nature of many of these bullshit jobs means that they are easily replaced by AI without the need for investment in machines or robots.

Is it theoretically possible to build an AI-powered robot that works as a plumber? Yes, but that technology is still years away. And more importantly, the economics of a robot plumber probably don't work out for many more years: The median salary for a plumber in the US today is just over $60k, so replacing him by a robot that costs millions just doesn't make sense. Amazon tried "just walk out" cashier-less AI technology in stores, and ended up ditching the technology and bringing back cashiers; and it turned out that part of the technology was actually outsourced to India, where humans instead of AI watched video feeds. As that happened with several different companies touting AI, the running gag is now that AI stands for "Actually Indians". Even where AI can work, it shifts jobs rather than totally replacing them: AI-generated legal documents are now double-checked against hallucinations by the same paralegal staff that used to be writing those documents. Nobody trusts AI to calculate the statics of a bridge or building without human supervision.

Given that at the same time there are predictions of economic doom due to a lack of qualified people and demographic decline, the prediction that AI will make everybody unemployed seem especially silly. That is not to say that there won't be problems, e.g. how to get an employee the skills needed to supervise an AI when the entry-level job that was replaced by the AI doesn't exist anymore. But economic theory and two centuries of human experience with automation make it extremely likely that people will be still employed in the future, and have more productive jobs.

Maybe the AI doom among social media content creators is due to the fact that many of them are working a bullshit jobs that can easily be replaced by AI. If you create content that only caters to the algorithm rather than being your own creative expression, an AI can probably do it better.

- posted by Tobold @ 9:22 AM Permanent Link (5) comments

Monday, June 02, 2025

Failure is relative

As any fan of the Civilization series of games, and you will hear that Civilization VII was a huge failure: On Steam the recent reviews are "mostly negative", and Steam user numbers tell us that on any given day both Civ 5 and Civ 6 have more players than Civ 7. But there is another side to that story, which isn't as obvious as it seems.

The boss of Take-Two, publisher of Civilization VII, last month said that "sales are strong", and confirmed that "We're really happy with how it's going". That isn't just corporate speak, the stock of Take-Two is up over 20% since pre-release on February 6. So, what happened? Previous iterations of Civilization were primarily PC games, with console versions existing, but often with less support, less frequent updates, and later releases. Civilization VII was on consoles on day 1, with much of the UI adapted to consoles (which is part of what the PC players hate about it). Civ 7 sold more copies on the Switch than on PC, and on the Playstation sales were even higher. Industry analysts estimate that only 17% of the sold copies of Civ 7 were on PC. Overall Civ 7 is currently the 8th best selling game of 2025 in the US, which is far from being a failure.

This shows that Steam is a platform which is surprisingly transparent with commercially relevant data of every single product they sell. Consoles are not. Every YouTuber can make a "data driven" video about the failure of Civ 7 on Steam, while the success of Civ 7 on consoles is mostly invisible to the public.

Commercially, making Civilization better on the console makes sense. To have had a "good" release on Steam, PC Civ 7 would just have had to sell two to three times more, while on consoles it sold six times more than on PC. But I am sure that Take-Two isn't happy that if you search for Civ 7 on YouTube, you get mostly videos of what a failure the game is. Maybe Firaxis should have made the two versions of the game on PC and consoles more different, instead of making a "one size fits all" game. A mouse and keyboard control is fundamentally different from a gamepad control, and PC gamers weren't happy with a simplified UI for gamepads which then lacked the more detailed information they were used to from previous UIs designed for mouse and keyboard.

On the other hand, maybe it is impossible to make a Civ game that makes both console and PC players happy. Or a Civ game that is attracting new players and satisfying for veterans. Paradox is going the exact opposite way with Europa Universalis 5, making it more detailed than EU4, and nobody dreams of making a new player friendly version that is playable with a gamepad. EU5 is a game made by Europa Universalis veterans for Europa Universalis veterans, and that limits it to PC. The crazy consequence is that if EU5 is in decent shape at launch, it will be better reviewed than Civ 7 was, while making significantly less money.

- posted by Tobold @ 8:40 AM Permanent Link (6) comments

Wednesday, May 28, 2025

Small Western Trail

Earlier this year I bought Great Western Trail: El Paso, and yesterday I got it to the table for the first time. I had studied the rules, but we were 4 players who never had played the game before. Setup, explanation, and a full 4-player first time playthrough took us 2.5 hours. Which was a good length for my typical board game nights, as we usually have a maximum of 3.5 hours, and we play different games with different people, so it is rare that everybody has already played the game before.

Since the original Great Western Trail from 2016, there have been a second edition, variants like Argentina or New Zealand, and an expansion called Rails to the North. All of these easily take 3 hours to play, which doesn't include rules explanation time and the added delay when playing with new players. Which is why I never played any of these games, they are simply a bit too long for my game nights. Experienced players who don't suffer from analysis paralysis can certainly play Great Western Trail in under 3 hours, but for my typical player round the risk would be high to get kicked out of the game store at closing time without having reached the end of the game.

Thus the big advantage of Great Western Trail: El Paso being shorter, and I very much enjoyed my first game. Having said that, I have mixed feelings about other aspects of the downscaling strategy. They made the game not only shorter, but also smaller, and cheaper. The game board is made out of cloth, leading to the first time of me ironing a game board, as it gets creased when folded in the box. The cards are smaller than in the original game, and the money tokens are so flimsy cardboard that I used some metal coins I had in reserve instead. While all this make El Paso also a lot cheaper than its bigger brothers, the result was a bit substandard for my tastes. I generally prefer board games with quality components, tokens, and cards.

In Civ7, Antoine de Saint-Exupery is quoted as saying: “An engineer has achieved perfection not when there is nothing left to add, but nothing left to take away”. While I haven't played the other Great Western Trail games, I am certain that they do have a deeper decision space. But getting to a very similar gameplay experience with fewer game mechanics and rules is an achievement. And I can think of several other board games which would benefit from the same treatment. If a game has a weight of over 4 on BoardGameGeek and takes 4+ hours to play, you could probably turn it into a shorter and more accessible game by taking things away. And if that is well done, the fundamental game experience can remain similar, just easier. Which makes the game accessible to more people, and could therefore improve sales. What's not to like?

Labels: Board Games

- posted by Tobold @ 12:53 PM Permanent Link (0) comments

‹Older