I'm in the middle of a suit against them for harassment received from surveillance employees. Can't say much more than that (active case), but it's extremely common. The methods used once surveillance is discovered are deliberately made to be so unusual that you're either terrified into silence or are afraid no one would believe you, so that the operation can continue.
The convergence is only valid if the distance ladder is accurate. There are a variety of deductive bottlenecks in the distance ladder which could implicate the whole current distance model. Standard candles and redshift measurements are calibrated together, for example. If either is off then the whole current ladder could be invalid.
Probably some combination of quantizing down from original fp16 weights and changes to the system prompt used for chat. Both can cause degraded quality, the former more than the latter.
I attended lectures of his at IU. My impression is that he has some sort of chip on his shoulder. He's been committed to the ideological position that AI is Hard and Won't Happen for decades, and having models be so competent at translation and other tasks (he used to argue that google translate was essentially impossible) has probably thrown him off balance.
So the refugees who put their crypto on a thumbdrive so they could get their money out of ukraine as the infrastructure was bombed out of existence are scum of the earth? There are absolutely 0 legitimate use cases?
Regardless they're IPOing, so it's not even about revenue or profit anymore. And it's not even about growth. Public companies are now about growth of growth and are evaluated on that metric. So the incentives suggest they're about to start an intense phase of milking/rent-seeking their users to extract as much value from their existing vertical before they have to make other parallel products (which seems like it's probably going to involve yet another tiktok clone, unfortunately).
I have a completely new install of archcraft with hyprland and I have to say that it's the fastest out-of-the-box productivity experience I've ever had.
Hmmm, I see that Stacks has 2 seed rounds, a series A, and a VC round all separately listed prior to the token fundraising. Good to see you were able to be partly successful at traditional raising, at least enough to afford the lawyers and the filing fees, before you had to raise from tokens. Not every one has those opportunities or connections.
You reach out to an agency (generally CFTC and/or SEC). An attorney from that agency eventually contacts your attorney. You have a conversation. You hear nothing back, so you reach out again. A different attorney reaches back out again months later and you rehash the same conversation again. Repeat for years. Best case scenario it goes nowhere and you do voluntary third party audit/compliance/monitoring to demonstrate good faith. Worst case scenario despite any good faith you get a notice because you were already on their radar and the newest attorney smelled blood instead of just wanting to talk again.
> How many cryptocurrency companies do you know of that could pass an audit?
This is, quite simply, zeitgeisty prejudice. Most well-funded cryptocurrency companies have passed audits. I've worked for or with four crypto companies and three of them were audited (also all of them still exist).
You aren't allowed to count "fraud cryptobros from miami creating an ICO" if you don't count "fraud fratbros from harvard/yale creating a hedge fund". And if you do count both, then you'll notice that the proportion of fraud is roughly the same everywhere, we just allow ourselves to notice it in some places more than others.
> Why should the SEC bend over backwards to create new rules
Because you don't apply old rules to new contexts if you're a responsible government, especially if you're claiming jurisdiction over those new contexts. Registration and filing processes created for the paper era simply don't cut it for regulating self-executing financial instruments. Why doesn't the SEC have a registration smart contract? It is the SEC's responsibility to adapt to the times.
> Would you kindly point out a company that tried and failed to register an ICO
What do you not understand about "there is not process"? I've been a part of two companies who have tried. We have had "conversations", aka we try to reach out and a different attorney occasionally gets back in touch with us and rehashes the same conversation every six months. The longest effort went nowhere in four years. There is no registration, there is no process, there are no boxes we can check on the existing forms which govern what we do.
That's literally how creating a theory works. You don't "pre-hoc" explanations or there would be no point to scientific inquiry. Do you mean to say he's cherry-picking or displaying some other bias?
It is his field. He's a geographer and a biophysicist, not only an ornithologist. Understanding geographic and environment contributions to evolution is explicitly in his wheelhouse.
I keep seeing this robot-like criticism of Diamond as "over-simplistic" or "hyperreductionist". But it's just a model? The map is not the territory, and Diamond has never claimed to explain all human development. His thesis is merely that geography and environment contribute to biological and cultural dominance to a disproportionate degree compared to other factors. I don't see how this is controversial.