1,798 karma · joined January 13, 2011
We talked about the state of reinforcement learning and what the field needs to succeed in building autonomous agents, the challenge of founding and building a company with a research-first agenda, which AI technical advances are over- and under-hyped, and much more.
Kanjun thinks language models are _still_ underrated!
Kanjun thinks language models are _still_ underrated!
1. Use some sort of instruction tuning to get the thing "good enough" that it gives decent results 75% of the time and the other 25% a human has to take over. 2. Use the actual usage data as training input. Punish bad behaviors and show the model what the human did to solve the problem. 3. Use this training loop to progressively have the model take over a larger % of the time.
…and I think if you can't get (1) good enough to be worth using it's going to be really hard to get the loop going.
First off, lots of other currencies are valuable without being reserve currencies. Swiss franc, Japanese yen, etc. Being a reserve currency probably boosts USD a bit but it's not the only pillar holding the thing up that people imply. If you disagree show your model: You have to believe that value for the euro, GBP, JPY, Swiss Franc is determined by thing X, USA does not have thing X and instead depends on being a reserve currency.
Second, what's the alternative reserve currency? Bitcoin? Don't make me laugh. RMB? Even more ridiculous. Euro? Probably the most realistic contender but seems to have major governance problems. As with democracy, the USD is the worst reserve currency, except for all the alternatives.
The government gave $18B to the SLS and so far has vapor: https://en.wikipedia.org/wiki/Space_Launch_System
The government spent $211B on the shuttle program and got 133 launches. SpaceX will probably surpass that number this year at a fraction of the cost.
NASA's record on rockets since Apollo has been abysmal.
I don't think government is necessarily bad (the Russians did a much better job in recent decades!) but it leans into its failures and often has bad incentives. SpaceX fails fast, has great incentives, and has achieved an incredible amount on a (comparatively) shoestring budget.
I'm not 100% sure what the FRED chart you cited means by "Effective Loan Rate for Small Business Administration" but those are most likely collateralized.
- Noise means you can't do US domestic
They don't seem to be targeting this.
- Concorde didn't have the range for Pacific
They do seem to be attempting Pacific range.
- Costs didn't work for Atlantic routes
They are trying to bring down costs considerably.
- And airlines want lots of identical planes, not one special one for one route
This will certainly be a drawback, although if they could take e.g. 50% of premium transoceanic it won't be so specialized.
If anyone's interested in actually learning about the causes of mass incarceration, I'd strongly recommend John Pfaff's _Locked In_: https://www.amazon.com/dp/B01L6SLKK8/ref=dp-kindle-redirect?...
In short: a) It's not the war on drugs. b) It's not private prisons. c) It's not sentencing laws.
You could get rid of all of those and America would still lead the "free" world in incarceration.
Inefficient and expensive systems can do harm, regardless of the good intention behind this law. Rooftop solar has substantial embedded carbon and impacts on the grid that could easily outweigh the small benefits. We need a solution that provides the right incentives to use it when it makes sense, and not use it when it doesn't make sense, rather than a hamfisted blanket mandate.
A requirement is a very expensive way of foisting a certain lifestyle on everyone, regardless of whether they want it or not. You can't just say "oh it would be nice," you have to actually justify why it's okay to force people to do things this way, and making housing much more expensive.
My wife and I live in San Francisco, don't have a car, don't have a garage. Are you saying I should be forced to pay for not just one, but actually two spots?
Square footage in San Francisco is extremely precious. If people aren't willing to pay for parking I don't see why it should exist, or why people without cars should be forced to pay for it.
To be clear, I'm against parking maximums as well. If people are willing to pay for a spot, let them do so!
…also, regardless of who invented it, it seems like this design feature is unavoidable. If you want to kill the bezel but still want to have things like front-facing cameras, you're pretty much stuck with a notch.
…is that also why India "succeeded"?
Liberalization is a matter of degree. You can't cherry pick a particular protectionist policy South Korea had and then say, "Oh look see they used import substitution and still had a good outcome." You have to also explain why that import substitution policy didn't ruin the Korean economy the way it did Argentina, or in an extreme example, Maoist China. And what you'll find is that they became much more liberal in other ways.
If you're going to claim that liberalism doesn't matter you need to show your work. You need to show that in aggregate, countries that liberalize their economies do less well than countries with less liberal economies. But when you look at the data, you will find just the opposite: Countries that move in a liberal direction, like China, South Korea, and Chile, tend to improve their outcomes, while those that do not, like Argentina or Zimbabwe, tend to stagnate. And almost every wealthy country in the world is either highly liberalized or sitting on an ocean of crude oil. The fact that even liberal countries have illiberal policies doesn't disprove the benefits of liberalization.
Most likely, Asia succeeded in spite of protectionism, not because of it.
Crypto actually proves that money doesn't need to be backed by anything. It simply achieves value by being a) useful for transactions and b) scarce. The US dollar's scarcity is carefully handled by the Federal Reserve which is why it has stable value. In Venezuela, the government simply prints money to spend without a thought to managing scarcity, and hyperinflation is the natural result. Collective belief is important, yes, but it's simply a belief about the future scarcity of the currency.
There's a completely different question you are raising of whether there are zero marginal product workers. One could ask if people are made ZMP by trade. Of course, there have always been ZMP workers (an extreme example might be someone who is quadripelegic and has mental health issues), but trade could still be a factor in increasing their number. I would say with a cursory empirical look, in the cases where one could make the argument (e.g., high unemployment countries in Europe), it looks a lot more like this is caused by poor local policies rather than by trade.
In general, the model is focused on production not full employment (which there are different models for), and it's not clear to me why we would optimize on a single KPI here.
Infamously, at least in the Bay Area, the "undesirable things" in question are typically new neighbors.
…looks like the market expects 1.57% inflation over the next 10 years, so these bonds are expected to beat inflation handily.