There was also related discussion about another longform piece by the same author that I'm too lazy to look up at the moment..
In my opinion, this author has drunken the kool-aid and then some. There is simply no evidence that more scaling of LLMs will lead to AGI, and on the contrary there is plenty of evidence that the current "gaps" that LLMs have are innate and unsolvable with just more scaling.
If the FDA didn't exist, how many more harmful drugs would there be? Would those harmful drugs that had approval retracted still be out there being prescribed?
Just because the FDA is not perfect does not mean that it's a net negative.
Easy to have the best of both worlds, though. FDA can continue to collect and publish data on safety and efficacy of medications which is incredibly useful and absolutely worth the tax money. But if you want to take something not approved, then they should have no right to tell you that you can't take the risk.
And if you get seriously sick from that we let you die by the wayside?
The first way to take advantage there is to outsource the cost of dangerous experiments to (sometimes) desperate people and then to society at large for picking up the pieces.
That's the hypothetical half of the equation. The tangible, measurable half is how many people the FDA has killed by denying access to life-saving treatments. A number of studies[0] have been conducted on this topic.
Right, of course. Any preventative treatment is difficult to justify, as it's a risk waylaid, whereas treating symptoms has a visible direct impact on problems faced today; nevermind that it costs 500x as much and has worse outcomes.
One of my favorite bits of Vinge's A Deepness in the Sky is the use of base-10 time: ksec, Msec, etc. There is a nice time log scale with Earth time to base-10 time conversions.
Yes! It is as a direct result of that book that I now know without having to look it up that a ksec is about a quarter hour and a Msec is on the order of a fortnight, which comes in handy when doing back-of-envelope estimation more often than you'd expect. (I'd already known that a Gsec was about a third of a century thanks to Tom Duff's observation.[0]) I don't see us moving to such a system anytime soon in general (tying to the circadian cycle is just too convenient) but I'm a little surprised I don't see it more often in discussions of humans in space.
[0] "How many seconds are there in a year? If I tell you there are 3.155 x 10^7, you won't even try to remember it. On the other hand, who could forget that, to within half a percent, pi seconds is a nanocentury." --Tom Duff
The Emergent "utopia" is both horrifying and eerily believable. I've known some grad students and some tech workers who are way too close already.
But it's reallly strange to try to map the Emergent political structure onto any modern political axis. It's not "liberal progressive" or "traditional conservative" or "libertarian". Or any other popular political ideology. It's certainly authoritarian, but uniquely so. It's almost a dystopia run by project managers and exploiting specialists.
Also a fun bit: The traders in the book count base their epoch on the first moon landing, but if you pay attention, the lowest levels of software count from a different epoch.
They merely got confused... thousands of years from now, they assume unix epoch time is based on the moon landing. And it's only a few months off anyway. Not much is left of Earth as far as they know, to be able to properly understand that those were two different events.
The Emergents are what they are, because they are, at the most fundamental level, busybodies who want to control others. Sometime in their own history, they found an excuse (the Emergency) to do that, and they never stopped doing it even after the crisis was over. In this way they map to most other authoritarian regimes in reality, but especially to the leftist authoritarian regimes. They hate "peddlers" after all, who sell things to others at fair prices and of their own free will. Not unlike the mutterings you see all over social media concerning capitalism.
- my niece, an 11 year old, told me she want to be a doctor, and I knew that would be impossible unless I get her out of that situation?
- I've watched the 9 year old be physically unable to write any letters _even when having a printed alphabet in front of her_, be unable to speak the alphabet, be unable to even sing it?
- the 14 year old is on the hook for watching these kids, and is stressed and hurting, but we can do nothing?
Your throwaway comments are not helpful to people in real, hurting situations.
A type system like what a programming languages have doesn't make sense for databases.
For inserts, sure (but then, it's straightforward to use the host language for that).
But once you move beyond simple cases, data changes over time, not atomically. It's also common for data to be distributed. Static type system can handle either of these things.
If your database can't go down for the duration of a migration (which will need more time now, to type-check all the data), and you don't have a way to ensure all clients are updated at the same time also, then you can't use static type checking like programming languages have.
I’ve used sqlx and SQLite for a lot of projects and sqlx’s macros have prevented me from inserting the wrong types in columns, but I agree that it’s not a viabile long term solution, especially if you don’t use sqlx macros in the first place which is not something you can always do.
On the contrary, I think that everything should be fine with Diesel since that leverages the rust ecosystem to manage types and migrations of the sqlite database.
