almostdeadguy

I think its fascinating how many people in tech think there's a clearly defined and agreed upon "right way" of using this technology that everyone knows and abides by. Paul Graham, for example: https://x.com/paulg/status/2058871512451412457

It's like we memory holed the last 20 years of social media that was supposed to be all upside; just democratic, global connectionism, empowerment, etc. I have too much exposure to people using AI in various, even sometimes subtle "wrong ways" to really agree.

Linguists centuries from now are going to wonder why everyone suddenly started writing like Paul Graham.

Running your code once or even dozens of times is not proof of correctness, see decades of software engineering practices around testing.

I'm sorry, but all you guys pretending like you don't have to review code now are going to get seriously burned by this. I also have to ask: do you think you provide any value to your company?

People keep saying this like it’s some meaningful point, but the reality is many people in different projects have a shared need for that code to work correctly, and there is a social proof involved in used open source libraries. That is why people look at downloads and dependent projects as heuristics of stability and correctness. That is not the case with (and cannot be obtained with) code authored by generative AI.

I'm less sure of the fact that ending subsidized token consumption (in isolation) will happen and change this. I think we've seen this play out before with other tech companies where discounting early use ends up entrenching demand and allowing the company to build larger and more efficient infrastructure.

I'm slightly _more_ convinced (still not all that strongly) that the rising cost of memory and chips, data center construction that gets outpaced by computing demand, increasing energy costs, and low switching costs for customers will force the model labs to make changes that increase the barrier to entry (either via higher pricing, more restrictive rate limiting, etc.). or force their customers into longer term commitments.

I think the counter-intuitive examples of people who attend that you and others in the replies are pointing out are a demonstration of how many contradictions exist in these principals.

I am the type of person who thinks many, many things about the way the world currently exists need to change, but I am incredibly skeptical of the purported mission of the Burning Man Project to "extend the culture" of these principles to the wider world.

These people in fact are some of the principal figures dictating the dominant culture and status quo.

Happy for everyone trying to invent SEO hacking for resumes.

That the industrial revolutions use of steel to augment or replace labor was similar in every way to using LLMs to do the same? Seems on point to me.

Famously Andrew Carnegie spent years trying to get the steel to stop talking about goblins.

To me that’s a hallmark of a game that isn’t very interesting! If you can discover a dominant strategy in a handful of plays, the game probably isn’t worth your time IMO.

I think there has been a shift in what kinds of games get published that privileges a slim set of experiences that are possible in board games and risks narrowing the range of what people think a board game is capable of. I agree there’s a vast range of different player types and psychological rewards people get out of playing games, but I personally find myself increasingly uninterested in new game designs, because the designs I like are harder to sell to impulsive buyers, players who don’t want to play a game repeatedly, or players who will have difficulty playing games again if anyone has a bad time (which I totally get! But it means designs that might prompt negative emotions are not sought by many publishers). I wouldn’t even say “heavy” games are the problem (I disagree with OP about high time commitment being a problem, there’s many games like that that deliver commensurate value to me).

If you are not yet a player of modern hobby board games and are interested to try some I recommend this method for picking some to try out:

https://boardgamegeek.com/search/boardgame?sort=rank&advsear...

These are the top ranked games on BGG from 1990 - 2007. Pick a few of these, try to pick games that are highly differentiated from each other by category and mechanisms on their BGG pages (also I recommend you pick ones with under ~2hrs expected play time to start). Read the rules for one, go to your local board game cafe with some friends and play it. Try to recruit friends willing to play a game more than once, even if they initially dislike it (hard ask, I know, but sometimes games only reveal themselves after repeated plays). If you are enjoying this, repeat this with a couple picks and try to determine what are the features of these games that you enjoy (may be shared mechanisms, but eventually I think you will come to a more philosophical understanding of what you enjoy about board games, if you enjoy them!). You may find out early in this, that you don't really enjoy games except for the social experience. That's extremely common.

I do not recommend looking at kickstarters or the current top ranked games on BGG, or looking at recent youtube reviews.

We're living in a renaissance in terms of the quantity of board games being released, but personally I am finding a declining number of titles I'm even remotely interested in each year, and many of the titles I am interested in are remakes/reprints of older titles!

I think with the increased public interest in hobby board games, the pursuit of profit has lead to some really negative changes in the games that tend to get published.

100% agree, I think the past 10-15 years of changes in the hobby have been profoundly negative.

- The trend towards videogame-ifying board games

- The trend towards "cozy" games, i.e. games that are not interactive, have no potential to produce negative emotions, and focus on a solitary optimization puzzle.

- The kickstarter-ification of games that focus on early release exclusives, excessive plastic, aesthetics over game design, etc.

I really urge players today to look at some of the games from the 90s to early 2000s if they're interested in getting into the hobby. Seek out some of the "classic" hobby games. Even some games predating that are fantastic, but you will also run into a lot of over-the-top simulationist war games during the 80s period.

Figure your site deserves a plug ;)

https://18xx.games/

- 18xx games (1889, 1830, The Old Prince 1871, etc.). Basically stock market games built around running train companies.

- Games by the publisher Splotter Spellen (i.e. Indonesia, Food Chain Magnate, etc.). Interactive games, usually with an economic bent. Turn order manipulation is a large part of these games. Splotter games often feel like they are designed in a lineage similar to Uwe Rosenberg games, where you can see threads of design traits shared between games.

- Carl Chudyk designed games (Innovation, Glory To Rome, etc.). Games that feel random and broken but have lots of tactical play embedded in them. Tempo is challenging to figure out in these games (IMO), and sometimes there is a non-linear progression aspect to them.

- Older euros, predating the trend toward solitary play: El Grande, Tigris and Euphrates, Bridges of Shangri-La, Medina, etc. No single connective feature, but these are games that are more on the combinatorial and strategic side but predate the development of the "personal player board" as the primary place the game is played.

- Pax Games: Pax Pamir, Pax Porfiriana, etc. History-based card tableau games that all feature a conveyer-belt market mechanism (where you buy cards from a market and cards get progressively discounted the longer they're visible and give you turn lookahead). Semi-economic, but more about the interaction of card abilities (and sometimes map play). Very fun weird games, just ignore the footnotes in games designed by Phil Eklund (I also don't love the futurist optimism in Matt Eklund's Pax Transhumanity, but that's me).

- Some abstracts (such a time investment to get deep into these, but they're obviously fantastic games): The Gipf Series, TwiXt, Hive, Paco Ŝako. I'm not yet sure what type of abstract games I most enjoy, still figuring that out.

I tend to like strategic, competitive games with higher interactivity, but with lower amounts of "take that"-type interactivity.

The implicit rationalization here is "this is a problem that pre-dates our involvement, everyone is already doing this". Classic dodge for bad actors.

There’s a funny tendency among AI enthusiasts to think any contrast to humans is analogy in disguise.

Putting aside malicious actors, the analogy here means benevolent actors could spend more time and money training AI models to behave pro-socially than than evolutionary pressures put on humanity. After all, they control the that optimization procedure! So we shouldn’t be able to point to examples of frontier models engaging in malicious behavior, right?

This was an excellent piece with many new pieces of information in it. Thanks to you and your coauthor for getting it released.

Seems this got buried from the front page very quickly

I have yet to meet anyone whose problem with AI is that the code is not aesthetically pleasing, but that would actually be an indicator to me that people are using these things responsibly.

My own two cents: there's an inherent tension with assistants and agents as productivity tools. The more you "let them rip", the higher the potential productivity benefits. And the less you will understand the outputs, or even if they built the "correct thing", which in many cases is something you can only crystalize an understanding about by doing the thing.

So I'm happy for all the people who don't care about code quality in terms of its aesthetic properties who are really enjoying the AI-era, that's great. But if your workload is not shifting from write-heavy to read-heavy, you inevitably will be responsible for a major outage or quality issue. And moreso, anyone like this should ask why anyone should feel the need to employ you for your services in the future, since your job amounts to "telling the LLM what to do and accepting it's output uncritically".

Yes, anthropic is not the only company in the world with some shitty code, and yet I feel no pangs of guilt over laughing about it.

There's a reputational filtering that happens when using dependencies. Stars, downloads, last release, who the developer is, etc.

Yeah we get supply chain attacks (like the axios thing today) with dependencies, but on the whole I think this is much safer than YOLO git-push-force-origin-main-ing some vibe-coded trash that nobody has ever run before.

I also think this isn't really true for the FAANGs, who ostensibly vendor and heavily review many of their dependencies because of the potential impacts they face from them being wrong. For us small potatoes I think "reviewing the code in your repository" is a common sense quality check.

A defining work of the "just vibes" era.

I think the unstated (but highly prevalent) view among executives in large swathes of this industry is that they don't really care to spend any time or money on user testing or quality assurance, and if this role exists at companies it is usually under-compensated and straddles both these functions to have some party be accountable. It is sometimes a check on product teams and vision-driven executive teams who don't prototype/test their ideas (or empower their teams to do so), and sometimes a check on engineers and engineering managers who don't want to be accountable to gaps in quality.

The pathfinder algorithm is a great example of why constraints are so important for creativity and creative development.

If AI has any benefit to creative endeavors at all it will be because of the challenges of coaxing a machine defined to produce an averaging of a large corpus of work (producing inherently mediocre slop) provides novel limitations, not because it makes art any more "accessible".

Odious is one of the most reserved words you could use to describe Telegram, which is primarily a host for scams that the influencers and other bottom feeders aren't allowed to monetize on the big social networks.

If we can't spend that much time reviewing code, what are we exactly doing with this AI stuff?

I don't disagree, I think reviewing is laborious, I just don't see how this causes any unintended consequences that aren't effectively baked into using an AI assistant.

I'm sorry what? Junior engineers can't learn anything without using AI assistants (or is the implication that having seniors review their code makes them incapable of learning?) and senior engineer would hate their jobs reviewing more code from their teammates? What reality do people live in now?