arXiv doesn't need much. All they do is host static pdfs uploaded by someone else with free CDN services from Fastly [0]. I'm sure they could get academics to volunteer moderation services as well.

In reality you could host the entire thing for well under $50k/year in hardware and storage if someone else is providing a free CDN. Their costs could be incredibly low.

But just like Wikipedia I see them very likely very quickly becoming a money hole that pretends to barely be kept afloat from donations. All when in reality whats actually happening is that its a ridiculous number of rent seekers managed to ride the coattails of being the defacto preprint server for AI papers to land themselves cushy Jobs at a place that spends 90+% of their money on flights and hotels and wages for their staff.

I'm already expecting their financial reports to look ridiculously headcount heavy with Personnel Expenses, Meetings and Travel blowing up. As well as the classic Wikipedia style we spend a ton of money in unclear costs [1].

Whats already sad is they stopped having a real broken down report that used to actually showed things. Like look at this beautiful screenshot of a excel sheet. Imagine if Wikipedia produced anything this clear. [2]

[0] https://blog.arxiv.org/2023/12/18/faster-arxiv-with-fastly/

[1] https://info.arxiv.org/about/reports/FY26_Budget_Public.pdf

[2] https://info.arxiv.org/about/reports/2020_arXiv_Budget.pdf

> arXiv doesn't need much. All they do is host static pdfs uploaded by someone else with free CDN services from Fastly [0]. I'm sure they could get academics to volunteer moderation services as well.

This just isn't true. arXiv nowadays has to deal with major moderation demands due to the influx of absolute drivel, spam, and slop that non-academics and less-than-quality academics have been uploading to the site.

Moderation for arXiv isn't perfect or comprehensive but they put so much work into trying to keep the worst of the content off their site. At this point while they aren't doing full blown peer review, they are putting a lot of work into providing first pass moderation that ensures the content in their academic categories is of at least some level of respectable academic quality.

Volunteer moderators are a valid option. And I think may work out better than paid employees.

volunteer moderators are a valid option however this is also the way peer review works and the system is unfortunately very problematic and exploitative.

First pass sanity checks are also a lot less fun than proper peer review so paying moderators to do it is probably safer in the long run or else you end up with cliques of moderators who only keep moderating out of spite/personal vendettas against certain groups or fields.

You can have locally zero sum effects.

Either you fight over a saturated market. (Zero sum)

Or you try to grow the market. (Non zero sum)

Or you try to diversify into new markets. (Mixed, non zero sum if its greenfield research into new technologies, zero sum if old).

I mean why would they.

If they lived in fantasy land with reasonable permitting and running a toilet ment paying for toilet paper, soap, and a janitor to clean it once a day I'm sure they would for the PR win and to sell a few extra Teslas.

In San Francisco as discussed in the article this would be an expensive permitting hassle, endless money sink, target for abuse, bad PR from people complaining about unsafe and unclean toilets, and a legal risk if any incidents happen.

Doing it well (so the toilets are clean and safe) obviously costs money. And obviously Tesla would prefer not to pay money. I don't think that's the issue you think it is. Tons of pollution mitigation efforts cost money that companies don't want to pay and yet work because of laws requiring them.

My city tried to force a public toilets ordinance on local businesses. Those businesses either ignored it or converted to take out only or other business models where they were not bound by the law.

It got rolled back pretty quick.

Plenty of businesses will simply call your bluff for such silliness. They do not exist to solve social issues local government created in the first place.

If local governments can’t even afford to keep public restrooms going at parks, public transit stations, etc. it’s not reasonable to expect businesses to shoulder that cost for them.

Tesla Superchargers aren't going to be able to convert to take out only or other business models

Such businesses will close shop.

Tesla superchargers make enough that they can cover the cost of bathrooms and still be profitable.

Keeping them clean is the easy part. Doing it well would require building the infrastructure, maintenance, manning the station, preventing drug abuse etc, and getting into liabilities due to said abuse.

Tons of businesses stop doing business if they are forced into efforts that dont make money. The city/state should own sanitation. Not private companies.

In my experience LLMs often have really solid insights in the thinking chains then vomit a nonsense score that doesn't make sense.

Now I'm not sure if this is actually an LLM only thing. Because I think people probably do similar when you ask them to give a number to things without providing a concrete grading rubric...

Single iteration waterfall is a broken process. You really need those late stage usage feedback signals unless your requirements were somehow captured by God.

> Single iteration waterfall is a broken process. You really need those late stage usage feedback signals unless your requirements were somehow captured by God.

Waterfall, as prescribed and as practiced, had feedback signals.

We already have/had recursive self improvement in technology. Its just mutually recursive self improvement.

And it will continue to be mutually self recursive because its not like the only input to AI is AI. Its a lot of things.

1996 is not now. This comparision makes little to no sense.

I'm sure if Apple provided support for installing your own OS on their M series laptops it would be incredibly popular. And I don't need to guess at this using weird 1996 research on microkernels because Asahi Linux exists and clearly there is interest in it.

Indeed, Apple from 1996 would not released Tahoe, most likely.

We don't need research because QNX, L4 and many others on embedded space do exist as well.

Do you forget what Apple in '96 was? Or are you saying that Tahoe is too polished for the Apple of '96?

Apple was not a bastion of quality in the 90's. They couldn't modernize the Mac OS, and that continued with little more than window dressing over what was released in the 80's. The Mac line up was a horrible mess of barely different models that needed a Ph.D to figure out what was different. The company was bleeding money and seriously close to bankruptcy.

The Apple of the mid 90's wishes it could release something like Tahoe.

Yes the 1996 Apple was on the edge of bankruptcy, yet Mac OS 8 was definitely much more polished than Tahoe.

Ya ok, unless you looked at it wrong, then it crashed.

OS 8 was a platinum theme over System 7. Which was a slightly better System 6, which wasn't significantly different than System 4.

System 7 was good for the time, OS 8 and 9 were not, and Apples inability to improve the OS were really starting to show. Windows 95 was a more stable OS than OS 8. Tahoe is better.

How many Tahoe related radar issues do you want?

Yeah - an OS that crashed every time you launched Netscape and you as an end user had to manually allocate memory to apps?

Not to mention that the OS itself was still mostly 68K emulated code even on PPC Macs and holding the mouse down over the menu caused all apps to stop running.

At least you could read the text and resize windows.

Apple circa 1996 would be charging for its updates and licensing out the software to Power Computing and UMAX. They were making a lot of "interesting" decisions.

They still are doing lots of interesting decisions, the difference is that now the piggy bank is full to care of them going badly.

Rip Blackberry Phone + QNX, you were so promising for such a short time.

Just because you can analyse it doesn't mean that it is better. Deep learning theory is unbelievably garbage compared to the empirical results.

In particular, please show me a worked example of a decision tree meta learning. Because its trivial to show this for DNNs.

Why do you believe poor people can't simultaneously be terrible people?

Have you never seen the kind of horrific rat race mentality that widespread generational poverty can drive in a society?

Like If you listed the poorest neighbourhoods in the poorest cities in your country do you get highly virtuous places filled with saints.

They don't. You invented that argument from whole cloth.

> it would show the ones with the least, being the most powerful and wealthy on this planet.

I think it should be logically inverted. Right now it says:

Not virtuous => Extreme wealth.

But I think it makes more sense as:

Extreme wealth => Not virtuous

> even after Donald Trump vowed to backstop trade through the key oil chokepoint

I mean you have to be completely insane If you take Trump's word for anything at this point. Pretty sure even people high up in the administration have given up on even pretending that anything he says makes any sense.