I'm still waiting for the "obvious scam that Tether is which is printing fake dollars as everybody knows" to implode. Any moment now.
BTW, wasn't the narrative that crypto goes up because it's bought with printed Tethers? Why aren't they printing some now to start another bull market?
Are you implying that Tether isn't shady? Because it's obviously a huge scam.
Now the question isn't if, it's when. And as Madoff showed us, it can take a very long time, decades even, before it fails.
And predicting the future is very hard, so nobody knows how much longer it'll take. But down it will go because it's inevitable.
And to your second paragraph, that's an oversimplification. Yes, printing Tether helped dampen the fall and prop up the price, but it did not do so alone and it's not as easy as just printing more to start the bull market.
Trying to do that would smore likely just hurry up their impending doom.
It is great as long as they are actually sticking to short dated US t-bills. They've just been super sketchy about if they are or are not in the past. Personally I think they're fine. The absolute worst case there's a run and depositors only get back like 75 cents on the dollar. Which is way better than most places that people park crypto.
My amateur theory is that Tether could have imploded for most of its existence, but now there’s a real chance that they have been (or will be) able to bluff their way to solvency.
You might just as well wait for the 'obvious scam of the FED printing fake money' to implode.
The thing with fiat money is, fake or not, as long as enough people believe in it, it functions well.
Replacing an imperfect implementation with one that has demonstrated time and time again it's far worse should never be the solution. Crypto has languished in obscurity (other than gambling, scams, and criminality) for 15 years because anyone who isn't in the cult (or hyping their "bags") can see how obviously ridiculous it is.
It's been clearly and obviously explained to death for crypto people where the value of fiat comes from and it's not a "debate" I'm interested in rehashing.
You are being downvoted but I think there's some truth. I'd like to see counter arguments. One that I can think of is the difference in intent. Tether is for profit and the FED is to stabilize the economy.
IIRC, the NYAG/JPM report from a few years back almost explicitly said "this is totally criminal and entirely fraudulent, but it's too late to stop, because if we do, the entire global financial system will explode"
I think you're remembering incorrectly. Crypto is large, but nowhere near large enough to cause contagion risk, especially because nobody is stupid enough to lever against it as though it was a AAA asset (like they did for American mortgages). If every single cryptocurrency was zeroed tomorrow, the global economy would shrug and move on.
Quite possible. The closest I can find is [1] (via [0]), which upon re-reading suggests the cryptoverse, not the global economy, would blow up. Perhaps I'm conflating it with the general warnings of the era [2,3,4].
Though I swear I recall reading some ominous wording, pointed out by another commenter, that subtly suggested the decision to quietly settle vs. aggressively prosecute was based on billions of dollars of potential economic fallout. Other articles [2,3,4] talk about contagion risk, but none are exactly what I recall. Funny how memory works!
One almost hopes it happens solely to shut up the crypto bros that still refuse to accept the primary use case for crypto is scams.
Sure it can be a distributed concencus/ledger system/smart contract system almost no one important needs or will use, or will reuse the idea internally themselves if it's useful.
Anything important won't be built on something they can't control or manage. Crypto by definition is that. So you have to really consider what this built on it and who's profiting from it.
Arguably the crypto blowup last year did result in some contagion, being one of the triggers for the US’s recent Medium Sized Bank Crisis. Though, also, arguably that was coming anyway, one way or another.
I think the 2023 bank crisis proves the opposite of what you're claiming. Only 3 banks failed, and of those three one had nothing to do with crypto. It was not a contagion, and people predicting it would have widespread repercussions on the economy turned out to be wrong.
"No contagion risk" doesn't mean nobody would be harmed by a crypto collapse; obviously there are a handful of morons here and there who would get wiped out. It means there is no systematic, widespread dependency of the real economy on the crypto one. I think the bank failures bear that out.
the term is contagion. and in this case it is probably hyperbolic, but basically when people rely on one asset as a medium of exchange, and that asset turns out to be of poor quality or worthless, then it affects many other assets and businesses.
if balance sheets and collateral for loans for big businesses were all in tether it stopped being redeemable or traded at $0, the businesses would have nothing on their balance sheet and all their lenders would realize the collateral was also missing, and the lenders also would realize they wouldn't get paid and had lent on bad assurances. The lenders would also lose money on all those loans, and their capital partners would lose money (private equity firms and their limited partners) and everyone that relied on payouts from the PE firms would have to change their forecasting, and the PE firm would not be investing in the economy anymore, making a hole in that market. Depending on how many PE firms were doing this it could grind a significant part of the economy to a halt.
good news is that typically the government fills in the gap. but people don't like that.
For function inlining specifically, I'm not sure LLMs are necessarily the right choice. The original MLGO paper [1] demonstrated a big code-size improvement with a ML model for making inlining decisions (7-20% code size wins), but they used tens of engineered features. Maybe a LLM could squeeze some additional size wins out, but maybe not [2].
Additionally, there are other factors to consider when productionizing these systems. Compile time is important (which LLMs will almost certainly explode), and anyone concerned about code size will probably be doing (Thin)LTO which would require feeding a lot more context into a LLM making inlining decisions.
I'm probably being very naive with this but, could mechanistic interpretability play a role here? Specifically, to experimentally short-list the LLM's most effective optimizations, then try to peek inside the LLM and perhaps fish out some novel optimization algorithm that could be efficiently implemented without the LLM?
I did a bit of work on this last summer on (much) smaller models [1] and it was briefly discussed towards the end of last year's MLGO panel [2]. For heuristic replacements specifically, you might be able to glean some things (or just use interpretable models like decision trees), but something like a neural network works fundamentally differently than the existing heuristics, so you probably wouldn't see most of the performance gains. For just tuning heuristics, the usual practice is to make most of the parameters configurable and then use something like bayesian optimization to try and find an optimal set, and this is sometimes done as a baseline in pieces of ML-in-compiler research.
Fentanyl needs to be made legal and sold at grocery stores. In 2-for-1 packs, fentanly+narcan, at affordable prices. Even better, subsidized prices for poor people which can't afford it.
That would solve the fentanyl problem, which is solely due to it being illegal and impure.
More generally, all drugs need to be decriminalized and sold at affordable prices and pharma-grade quality.
I used to share a similar position until I realized: drugs are bad, actually, and there should be substantial friction in making them available to people.
Disincentivizing drug consumption is a good thing. The war on drugs obviously leads to some absurdities - e.g. drug cartels, unnecessary incarceration - and I much prefer the Portuguese model.
But making fentanyl in particular available in the grocery store would be bad; some substantial number of people would die who wouldn't have died otherwise. Some substantial number of people who would never have tried fentanyl would give it a try.
Some sensible balance needs to be struck between a free society and preventing people from e.g. leaving live landmines in their parking lots. Doing things which inevitably kill people/mess up their lives without active violence is still bad, and should be heavily discouraged.
What led to your realization that "drugs are bad", and does that include all external consciousness altering substances? (caffeine, alcohol, nicotine, THC, and so on).
Can't speak for somebody else's post, but addiction ruins lives, not only yours but also possibly the people around you, and leads people to other criminal behavior.
It's perfectly legal to obtain and use alcohol for example, but alcoholism is an insidious evil disease.
Sure, yet you'll probably find no-one who is destroying lives (their own or others) for caffeine. THC is certainly a more mixed bag, but at the very least you will never die from it. So "drugs are bad" is just too much of a blanket term, and we should probably come up with better terminology that correctly reflects addiction potential and intrinsic and extrinsic damage. In addition, the illegal market shifts those dangers quite significantyl. Someone can take morphine for the rest of their lives with out much of any bodily harm, but in an illegal market it can be deadly quickly and procurement leads to much pain and harm. Last but not least, humanity will never not want to use drugs, no matter what we do, so we need to come up with harm reduction that is effective (which includes regulation, education and probably also includes doing away with prohibition).
yeah, definitely. "Drugs are bad, actually" is decidedly an over simplification. For the record my drugs of choice are SSRIs, caffeine, and mushrooms.
Full agree with everything you're saying here. The less oversimplified realization I had: In the pros/cons of "Should drugs be fully legalized?" I was counting only the negative aspects of making drugs illegal, not the negative aspects of legalization. I expect others make this error too, so "drugs are bad, actually" is intended as a pithy corrective, less a broad ethical directive.
The specific causes of this realization: the accruing evidence of the downsides of THC, and the easy access of teenagers to extremely potent weed. The ditch weed I bought from Curtis behind The Globe coffee shop in high school simply had less potential for harm.
> we should probably come up with better terminology that correctly reflects addiction potential and intrinsic and extrinsic damage
Full agree. Really there are complex policy trade offs here. I don't pretend to know optimal solutions, and agree that arriving at optimal solutions is unlikely via an extremely polarized discourse.
Drugs that kill or otherwise destroy your life are very bad. Other drugs that don’t do that are probably much less bad, and we should allow things somewhere that fall under a line between not bad and very bad. Where that line exists is a matter of current debate.
Making all drugs legal to use makes sense to me. Criminalizing users only makes things worse. But make it illegal to sell, transport, possess in large quantities etc.
The problem I see with criminalizing drugs (including sell, posses and transport) is that drug addicts and abuse is mainly a health issue.
There's an underlying reason why those people start consuming that shit. Same as with sugar/obesity/diabetes here in Mexico.
Once you make it ilegal, suddenly you created a criminal issue. You've created another problem for yourself and split your resources to fight both problems now.
I would rather prefer government/society focus only in the underlying health issue. And disappear the criminal issue by making it 100% legal.
Imagine if suddenly refined sugar became illegal in Mexico, suddenly the government would have to fight not only the obesity epidemic, but also the sugar trafficking criminals.
There are test kits available that can show you what you have, including tests for fentanyl. It requires a bunch of different reagent tests if you bought a Mystery Powder that you want to ID. But if you were after MDMA, for example, and bought what’s claimed to be MDMA, you can test it with a specific reagent to confirm whether it is that or not, then test it with a fentanyl test strip to show if it’s free of that contaminant.
Dance Safe sets up tents at big festivals to do this kind of testing as a harm reduction service. They (and others) are trying to normalize the step of testing what you have before using it, and I think that’s a wise choice especially with fentanyl in the mix. I’d expect that with decriminalized use of “white powder” drugs, you’d see these kits become more readily available from smoke shops, maybe even pharmacies, instead of being something you have to search for online.
Decriminalizing drug use is intended to solve a different problem - that being throwing people in jail for being addicts is bad in a number of ways and consequences.
Making all drugs legal and available (with quality assurance) will (at least help) solve the problem you are talking about, but will at the same time create whole new problems that more and easier access to drugs will mean more people try and use them and then an increase in lives destroyed through addiction (and related consequences). My opinion is that would be worse.
Decriminalizing use basically decriminalizes selling as far as I can tell here in the northwest. Police and prosecutors can no longer threaten users with legal action to give up their dealers, and since buying is now low risk, dealing also becomes lower risk. I’ve seen people on the street looking for drugs and find them on my morning walks (they are pretty vocal when asking, and dealers aren’t shy about replying).
I don't disagree. At your local grocery store you should have a drug aisle with cocaine, heroin, fentanyl, MDMA, mushrooms, all pharma-grade quality. So you would know exactly what you are getting.
