You don't think a $1000 gamble on this new paradigm of blockchain crypto was a sound decision? When the sum I was putting in was otherwise an insignificant sum to me.
I bought in fully knowing it could go to 0, or maybe it could be worth a ton in 10+ years. To me it seemed like the chance it would blow up was well worth the tiny risk of losing a pretty meaningless amount of money to me.
I am not even otherwise a gambler. I have never gambled at a casino or on sports or anything like that. And my stock investing is all index funds. This was the only singular "crazy gamble" I had ever made and I knew full well it was crazy. But the potential in my mind around the tech and the potential hype around it seemed to greatly outweigh the tiny risk.
You had a positive outcome, but yes, despite that, I don't know if that was based on a sound decision. It's possible to vastly misjudge the expected value of a trade and still come out ahead.
> This was the only singular "crazy gamble" I had ever made and I knew full well it was crazy.
The only thing that matters is whether it's positive EV (and whether your methodology of coming up with the EV itself is sound). If you didn't have any explicit or implicit notion of the EV at the time you made it... It was probably not a sound investment decision, despite being profitable.
I mean I thought that there was a potential in blockchain tech back when I bought it. I also thought that there was also potential in the hype around blockchain to explode the value simply from hype and how humans behave alone.
At least WAY more of a chance than what I would get spending $1000 on a lottery ticket or at a casino.
The entire crypto ecosystem is hardly all about deflation these days. If anything, I'd argue the opposite. Stablecoins, yield, perpetual futures etc. are hardly what Satoshi had in mind.
Humans writing tests can only help against some subset of all problems that can happen with incompetent or misaligned LLMs. For example, they can game human-written and LLM-written tests just the same.
As long as there's no moat (and arguably current LLM inference APIs are far from having one), it arguably doesn't really matter what users pay by.
The only thing I care about are whether the answer helps me out and how much I paid for it, whether it took the model a million tokens or one to get to it.
Clarifying ambiguity in questions before dedicating more resources to search and reasoning about the answer seems both essential and almost trivial to elicit via RLHF.
I'd be surprised if you can't already make current models behave like that with an appropriate system prompt.
Great video, but coincidentally two weeks too late for me:
I never knew about the boil auto-shut-off mechanism and the overheat/"boil-dry" protection circuit being two different things in some kettles. In those, pouring out the entire water before the auto-off had a chance to kick in can cause a situation where the kettle stays on (due to not enough steam being present to trip it), but the boil-dry protection circuit continuously engages and disengages until somebody notices or the thing self-destroys – ours did the latter.
Now we have one with a switch at the bottom, and I'm hoping that due to that construction, the boil-dry protection will also disengage the switch if needed. (It also helps that the switch automatically disengages when lifting the kettle off its base.)
I have a kettle with 5°-settable temperature targets that has transformed my tea drinking.
It also has a function to hold temperature for up to 30 minutes, and because it has actual logic going on inside, when you lift it off the base it knows this and won't turn back on when you put it back.
I have an electric water boiler like that also. I have to be very mindful to press a button to turn it off if i return it empty to the induction plate/holder otherwise it starts clicking and beeping and freaking out.
> If everything goes offline for one hour per year at the same time, then a person is blocked and unproductive for an hour per year.
The consequence of some services being offline is much, much worse than a person (or a billion) being bored in front of a screen.
Sure, it’s arguably not Cloudflares fault that these services are cloud-dependent in the first place, but even if service just degrades somewhat gracefully in an ideal case, that’s a lot of global clustering of a lot of exceptional system behavior.
Or another analogy: Every person probably passes out for a few minutes in their live at one point or another. Yet I wouldn’t want to imagine what happens if everybody got that over with at the very same time without warning…
Not too long ago, critical avionics were programmed by different software developers and the software was run on different hardware architectures, produced by different manufacturers. These heterogeneous systems produced combined control outputs via a quorum architecture – all in a single airplane.
Now half of the global economy seems to run on same service provider, it seems…
Netflix outside of the US is a very different experience.
In the US, it's mostly their own productions and older content they explicitly acquired, but elsewhere, especially in markets that don't have a local HBO or Disney streaming service, they have incredible backlogs.
I remember finding basically everything I could wish for on there when traveling in SE Asia almost a decade ago, compared to a still decent offering in Western Europe, and mostly cobwebs in the US.
reply