jmyeet

I've had that same thought. It's a surprisingly valid point even though people tend to think you're joking. A human is capable of consuming a certain percentage of all human output, admittedly a low percentage, but that might change with, say, vastly extended lifespans in the future. What percentage is allowed?

The first difference is that somebody is typically getting paid for any content you consume. You buy books, TV shows, movies, magazines, whatever. Or you borrow them from a library who bought them. Or you pay a subscription for a streaming service. Or you use an ad-supported platform. If you bought any commercial IP and then reproduced it one million times so you can train a whole bunch of people at once, well that's a crime. So why isn't it a crime when AI companies do it?

The second is that you can only produce so much output. You can't be cloned a million times to produce a million times the output. Software can be.

LLMs are kind of like a billion people sharing one NYT account, 1 Netflix account and so on. And that's simply not sustainable.

I also believe that the lifespan of IP is way too long. Software shouldn't be patentable because it's ultimately math and math formulae aren't patentable. Other patents should generally have much shorter lives. Books, music, TV, movies should have much shorter copyright periods, say 10-30 years total.

AI companies ingesting all the IP on earth for free is like saying there is no IP at all. Some will agree with that. I don't. But again, I would have less of an issue with it if we as a society owned the output.

I disagree. I believe these LLMs have already been commoditized.

Consider the primary use case for Anthropic: coding. Well, if you're using Claude Code or some variant, you can easily use OpenRouter to simply use whatever free tokens you can get. You may not even care what LLM you're using under the hood for a given task.

That's simply not a trillion dollar moat.

Can Anthropic jack up the prices? No. There are alternatives. There are open weight models that you can run on cloud platforms and increasingly you can run them locally. Local LLMs are going to be incredibly accessible in 3-5 years. Even now you can run Flash models on something you can buy from Apple. The inference is a lot slower compared to cloud usage but that'll change.

I see the Anthropic IPO as an effort to cash out before it all comes crashing down.

What AI is doing is socializing the costs but privatizing the profits. This has been the model for every social media platform going back to forums. Everything after USENET, really.

If you're using the collective output of humanity to train your LLM then humanity should own the output. Problem solved.

One can easily understand this as the worker's relationship to the means of production.

I really can't believe there's anyone left who thinks deregulation solves anything. Everywhere it's tried it is a demonstrable failure. It's just an excuse for an uptick in profits at the expense of safety, the environment, etc.

Safety regulations are written in blood. They exist because people died because the industries could not and cannot be trusted to regulate themselves.

The most generous interpretation is that people conflate regulation with rent-seeking. These are two entirely different things.

This is happening to SWEs too. If you have a certain amount of experience, you are capable of critically analyzing AI output. In Big Tech parlance, these are the Senior SWEs and higher. You require little to no supervision and can work independently.

But how do you make Senior+ SWEs? By having a pipeline of juniors who get trained and mentored. That pipeline is cooked now.

There is going to be a brief period where some like Staff+ SWEs are going to make a lot of money by being AI specialists. But the flood waters are rising for everyone. Those opportunities are going to shrink.

It's eventually going to be a problem that the ecosystem has died because there's no new blood and the people there lack the experience and critical thinking to understand what's being done.

I can use AI on topics I know because I can spot the errors. LLMs lie a lot and get things wrong a lot. What happens when nobody can check the output?

One of the things I was forced to reckon with over the pandemic was that I'd spent my entire life with undiagnosed autism. Looking back, there were lots of warning signs but we know things now we didn't then. For example, up until the 2010s the medical consensus was you couldn't have ADHD and autism.

One of the ways this manifested was my ability to anticipate problems at work and poke holes in things. This made me incredibly unpopular. You never get thanked for this. Instead you get punished for being right too early. You are negative. You are a naysayer. You aren't a team player. And you were 100% correct and it didn't matter.

Back in 2012 I remember thinking "I don't see how crypto has any applications other than illegal stuff". And I stood by that. Note that morality and legality are two different things. Sending money to relatives in a country that disallows foreign currency transactions may be moral but it's still illegal.

What's frustrating is dealing with the same consequences of being right too early. It's muddied by true disruption. An example that was often brought up was Google. In the late 1990s, it was believed that search was a "solved" problem and there was no money in it. That's 100% true. But you can't ignore any criticism with "people once thought the world was round". It's a useless fallacy to deploy.

The absolutely dumbest people and outright scammers were attracted to crypto. A bunch of people who missed out on Bitcon's meteoric rise didn't want to miss the next meteor or they simply wanted to rug pull before they themselves got rug pulled.

The financial system is an endless series of Chesterton's Fences. So many things work the way they do because of a problem that occurred. They say that safety regulations are written in blood. People died to make those regulations. Well, financial regulations are the same way.

Being able to freeze assets and reverse transactions, for example, is a feature, not a bug.

The real story with crypto is how greed clouds people's judgment. So many people on HN thought they were going to make their bag on crypto and people like myself got shouted down as naysayers.

I really don't know what to do with this information other than those same warnings are beign ignored about AI and in 5 years I will be punished for being right too early. There's simply no market to recoup the trillions invested in rapidly depreciating assets around software with no moat.

I love the Mamdani DoorDash story because it demonstrates that government can work. When it doesn't, it's because it's made not to work or the people in charge have absolutely no interest in doing anything than feathering their own nests.

Back during Covid, Congress passed and Biden signed the Inflation Reduction Act ("IRA"), which actually did a couple of good things.

One of those was the Corporate Alternative Minimum Tax ("CAMT"). For non-Americans, AMTs are a weird thing where Congress basically gives up and decides the tax code is too complex so they give up and say whatever your deductions are, if the AMT produces a higher amount, that's what you have to pay. Every complexity is somebody's special interest so repealing them can be hard. It's kind of why the carried interest tax credit has never been repealed.

Repealing it, as Republicans tried to do, is hard to do because the counter-messaging is "well if the corporate tax rate is 21%, why do you care about repealing a minimum of 15%?". Everybody understands that companies through various tax schemes aren't paying 21% but you can't say that.

But the one I want to talk about is the 10 year plan to increase IRS funding to update incredibly outdated systems and increase enforcement. This was a good measure. And you can tell because the Republican Party lost their minds. They fairly successively propagandized it as the IRS hiring 80,000 agents to go after Joe Q. Citizen. No, it was enforcement against the ultra-wealthy and corporations. So it was killed in 3 years under the guise of "saving money" as well as not persecuting average citizens (which it was never going to do).

It's particularly stupid because money spent on IRS enforcement pays for itself. It's been estimated every $1 spent recovers $7. It pays for itself. But there was no counter-messaging and Republicans could push it through under reconiliation.

Compare that to the 4 year $175 billion increase in ICE funding.

I don't know why people in these comments are dragging politics into it. That strikes me as a terminally online take and I say that as an self-confessed politics frog.

If anything, the following of current events overall is way down to what it was decades ago by the general populace. People overall are incredibly ill-informed about so many things despite the wealth of information available.

Three things happened to baseball (IMHO):

1. The childhood funnel is drying up and has been for a long time. Kids now increasingly play soccer instead;

2. Take out the nostalgia and you're not left with much. There are too many games, they're too long and not enough happens. Football is bigger than ever; and

3. Structurally I think baseball made some huge mistakes, such as the soft salary cap system that created dynasties like the New York Yankees who could get rich on broadcast rights for a huge media market. The NFL has managed all of this so much better.

I'm not a huge sports fan but I find the systems fascinating. American sports evolved into a 3 pillar system. Basketball season into baseball season into football season through the calendar year. I wonder if baseball's timing over summer has contributed to this decline.