conjectures

> A model will generate its own input and will watch it's own output and in process, will become more intelligent than it really is.

You're aware this is basically how AlphaGo was trained - playing against itself?

A bold move. I like it.

Not necessarily. Another scenario would be: Tesla anticipates the 2nd hand market getting flooded, undermining primary sales.

I think in many industrial settings Bayesian optimisation is more practical. Since it allows use to refine our design on the fly and put resources into promising areas of the space earlier. Some kind of pre-planned fractional design could be useful for generating an initial set of points to test.

Yeah the previous salary thing is a tough one.

It's also tough when it's reframed in terms of 'what are your expectations'.

I was wondering what the best way to field those questions is.

Puts an equation front and centre. Eschews brackets.

Amazon own brand jet engine, 4.5 stars, prime available.

I a fellow Brit, I note that you still bring up the war! But yeah, this is true.

Honestly, I don't think we get out of this hole until demographic churn (ugly phrase yeah) rebalances the electorate.

If we look at the 1992 election there's a 13% gap in conservative votes between the youngest and oldest brackets [1]. By 2019 that's a 36-46% gap [2].

[1] https://www.ipsos.com/en-uk/how-britain-voted-1992 [2] https://yougov.co.uk/politics/articles/26925-how-britain-vot...

It is depressing, but is it untrue? Yes, the prophecy can be self-fulfilling. But we have an impressive recent record of pulling the trigger on foot-guns.

op's comment included work. Unless it's common to measure social acquaintances by seniority.

Also, the existence of such a law should have a chilling effect on this behaviour. Even if it doesn't get rid of it. E.g. presumably HR policies & training will have to be updated to highlight it on pain of exposing employers to legal action.

:) why_not_both.gif

It's 2023 and apparently neither of the two main whiteboarding tools allow you to freely choose background colour. So forgive me for not being excited for either.

Keep things simple? Yeah, meet your new standard 6 key board, who needs the other 20 letters anyway: https://m.media-amazon.com/images/I/61C8GdWvptL._AC_SL1500_....

You're quite right. The true value could be much lower on average than the average self report held it to be. I hadn't considered that Graeber might have under-reported this phenomena if people are consistently over-optimistic.

Or maybe, absent other evidence, it might make more sense to tentatively take people's self reports at face value. Or at least hold off backing some particular elaborate alternative which involves people being wrong about the things they have most information on.

Yes, this is thread is developing some weird capitalist inverse of marxist 'false consciousness' theory.

Someone comes to you and says, "my job is pointless and should not exist."

The conclusion is drawn, "Actually their job is vital and their self-report is deluded."

C'mon that extra year makes all the difference ;)

To test the hypothesis that China is worse than the US, it doesn't suffice to point out that 'US bad'.

An exhortation to consider effect sizes is different to a claim about the effect sizes in a specific situation. It could be that OP is doing both, but you appear to be neglecting the former.

Yeah, I'm not sure why people are surprised that having a safety net allows poorer people to take more risks.

Do explain.

One really noticeable thing is the shrinkflation. Packets for everything getting smaller in supermarkets.

Actually I sympathise with this response. Getting tired myself of people larping as fact checkers and professional skeptics.

Not enough detail in someone's response? Here are some ideas:

1. Ask moar polite. Not, "I don't think the story adds up."

2. Assume the story is yet true and go to prove it to your satisfaction. Post the links and thank op for the interesting nerd snipe.

3. Enter the proposition into your mental database with some lower level of confidence and move on.

Yeah, tbh 'friends' are people for whom, when good stuff happens to them I feel happy about that.

Yeah, countries you can work in is pretty important here (even if slightly contrary to some of the spirit of remote).

Pour one out for the Machine God.

HN has its own groupthink. For crypto it's that it's all coal powered shiba shitcoins.

There was likely a similar group of people bitching in 1999. The dotcom bubble did happen, and yet dot coms also happened.

The point is that to see the upside, you need to evaluate what new capabilities the tech offers, whether they're interesting, and who is doing the real work (if it exists).

Moloch cares not for trees and flowers.

Yeah, there is a 'just change jobs' crowd who pop up in every conversation about working conditions.

There are switching costs which mean the employee often takes a hit on attempting to move. It's like buying a car, car doesn't work as advertised so someone says, "well stop whining and sell it, you're in market." Like yes, but also nope.

Meanwhile (probably elsewhere on HN), someone is hating on web3. Which might enable reselling all this junk.

> Avoid overriding, inheritance and implicit smartness as much as possible. Write pure functions.

Amen, preach it from the mountain.

Can't say I agree performance is bottom of the priority pile. As someone who does numerical work semi regularly, performance can radically change the usefulness of software (interactive versus batch for example).

One thing I'd vouch for was the improvement in my RSI/hand aches I got from switching to a £100 mechanical keyboard with mx speed silver switches (low activation force). Absolute godsend as far as I'm concerned.