Why not take it a step further? Make each function in the codebase its own project. Then the codebase can fit into the context window easily. All you have to do is debug issues between functions calling each other.
I don't think it's a joke about left-pad, but the idea that the complexity increases tremendously when you take a cloud of "small" things all communicating with each other. You've just pushed the complexity elsewhere. Claude can easily crunch the small microservice, but you're pushing the complexity to communications issues, race conditions, etc.
I switched to Zed for the first time over the weekend on a somewhat complex mixed C/rust project. I was able to set the whole thing up in about an hour to my liking and it is a really nice IDE, coming from bloated VS Code. I think they have a really nice AI-assisted coding setup, I think that the "file review pane, in line with IDE" UX is correct for AI tools. I'm skeptical that terminal or "agent" based AI programming is viable long-term.
Very cool. I think what's so remarkable is that this is basically the (ultra crappy *) version of the original hypermedia.
A person who knows something puts up a page, and uses hyperlinks to link to other pages. Those other pages have information from other people who are sharing their knowledge.
Of course, that isn't how the internet works now. Everything is a platform/app that wants to maximally surveil and control their users as much as possible.
* Not talking about the work of the person who made it; it's very cool and great UX. Moreso that the models themselves are an expensive, smeared, inaccurate/dishonest approximate version of actual human experience/expertise.
I mean, yes? I think when we say, we should force 18 year olds to do something useful for society instead of working doordash or taking college classes that they don't care about, that is kind of saying something about the usefulness of those things.
If the market figured it out we wouldn't be having these discussions in the first place.
It's not a market failure. If you pay enough money you can attract either citizens or even destitute African people to volunteer to get blown up for the glory of Trump and the oil companies.
I was first made aware of GPT2 from reading Gwern -- "huh, that sounds interesting" -- but really didn't start really reading model output until I saw this subreddit:
Certainly correct, but I think you’re underselling the historical exchange part of this. Dollars being everywhere causes the financial infrastructure to be built out in dollar terms.
Part of what enabled that huge capital flow you’re talking about is that it was the Americans who came in and gave [country’s] banks a counterparty to exchange dollars for oil.
A lot of that soft power is not just the ability of America to print dollars, but also the ability of America to control the financial infrastructure. To surveil, KYC, sanction, etc. that is a huge part of it.
The petrodollar is less mechanically important today but back in the day it was huge to have “everyone who needs oil” be the counterparty to a currency exchange. It is what injected all that liquidity, which set the whole thing off.
I think what people are realizing and considering now is with the computerization of everything, capital can flow more freely. That is what is dangerous (for the US) about today’s moment; our political leaders are taking it all for granted.
I do think history is also important, but again it boils down "where is a safe place to store my money?". That really controls everything else.
Now, in the past we had a gold standard, so you could literally move your money from one country to another. Now during both WW1 and especially the runup to WW2, the wealthy moved much of their money to the United States as a safe harbor, since we were the only advanced economy with deep liquid bond markets, rule of law, and foreign investment rights (sorry, Canada, but it's true).
This was the greatest wealth transfer in history. By 1940, the US held 80% of the world's global gold reserves. 80%! And this was in the era when international trade was settled in gold.
So it all happened in single decade between 1930 and 1940, and the US instantly became the world's global reserve leader, an extremely dominant position, merely because people were afraid of war and wanted a safe place to park their money.
After the devastation after WW2, the flood of European money into the US continued and more than offset the Marshall plan.
So already at the end of WW2, the majority of the world's liquid savings was tucked away in America.
Now, people like to tell stories of American soldiers spending dollars somehow making the dollar a reserve currency, and those are the types of things that seem plausible to people who don't monitor global capital flows, but that's honestly a ridiculous story. That was chump change.
Bottom line, there are no special technical reasons beyond "I want a safe place to store my money". That controls everything else.
There is an adage in the world of money markets: "It does not matter what currency you trade in, what matters is what currency you store the proceeds in".
And the moment that some other nation opens its doors to foreign capital inflow, establishes rule of law (which takes decades to develop a reputation for stability and not confiscating assets), is safe, stable, and secure, establishes financial transparency, and has deep, liquid capital markets -- then the world's wealthy will flood that nation with money also. But unlike declaring that "I will sell my oil for euros", doing the above takes decades of building trust and reputation. Gimmicks aren't going to do it when you are looking for a safe place to store your money.
Huh? The research done to develop the flight control computer for Apollo (and IBCMs of the time) lead directly to modern microcomputers. It’s hard to name something more impactful than that.
It could easily have taken another decade or two to develop the modern computer if not for the resources spent in the space program at that time. It still would have happened, but Apollo and the space program was soaking up something like 90% of computer demand for a full decade. Computers went from room sized behemoths to the size of a file cabinet in that time.
Im not sure that's an honest rhetoric, we have seen many other things in the last few years that have increased the demand for compute. It would seem lunacy to propose, to accelerate the miniaturization of compute we need to send a bunch of people to bounce around the moon, then we can forget about the space nonsense.
If the goal was begin the path that leads humans into so many resources it would take centuries before fighting over something was more profit than going to the next empty rock, we clearly failed.
reply