endymi0n

If you look at how the predecessor InstructGPT is trained, this doesn't look like a bias in political viewpoint, but rather a bias in the truthfulness of that very position. The fine-tuning step basically ensembles answer possibilities, amplifying viewpoints reflecting majority opinions and scientific logic, while rejecting fringe opinions, contradictions and logical fallacies.

It's pretty well proven the political right majorly prioritizes following authority and advocating partially contradictory belief systems over self-consistency, truthfulness and evidence based reasoning in most debates: From climate science over the war on drugs to private prisons, just to name a few uncontroversially debunked ones.

Apart from a few notable omissions on the left like a lot of really questionable gender "science" papers or political rejection of the greater male variability hypothesis, viewpoints on the left are on average far more self-consistent and based in reality than the ones on the right.

That alone imho explains a lot of perceived bias, which is the kind of bias I prefer: None but reasoning by first principles.

It might help the case here that apparently the model's "ego" is trained on helpfulness towards others, which might be a trained-in bias towards a more leftist stance.

I'll take that any time though over a model that's trained towards sociopathic behavior and subtle manipulation.

Does anyone have good examples of pre-product, pre-revenue companies going public actually working out, especially more recently?

I know access to capital is a big factor in this decision, but especially post Sarbanes-Oxley I have the feeling the forces against finding product-market fit are so incredibly high for public companies. Having seen first hand what that really means: The implicit taxes on morale of regulation, compliance, disclosure duties, transparency, auditing, bureaucracy etc. are so gigantic that I have a really hard time imagining the viability of this. I've found it's pretty hard to attract high freedom high responsibility talker-doers to a public company for those reasons.

That's certainly a driving factor behind startups staying private for longer than ever and continuously rising VC round amounts, but I already found it's hard enough for a profitable company with product-market fit to go public and still keep their top talent that I don't want to imagine what that feels like pre-revenue.

[citation needed]

Just looking at any of the commonly cited reasons to prep (unrest, natural disasters) especially in developed countries and how they played out in the end, what you see is neighbors banding together, volunteers arriving from all over the country and random people donating large chunks of money.

I like to think of the preppers‘ cynical and pessimistic mindset as ultimate projection: Just because they would defend their food stash with a gun doesn‘t mean others wouldn‘t share their last can.

Looking back, I guess the main point of why the early internet was so special wasn‘t because of cool new communications technology, but because of the self selection. In order to use BBSes or the early Internet, by definition you had to be above average curious, smart, rich, young (late night dialup tarrifs) and/or based in academia, as well as having a volunteering/cooperating mindset.

We just managed to keep the others out for a miraculous decade by keeping entry barriers hard, costly and complex. Is that an ableist viewpoint? No clue, but it was a bliss for sure.

I‘ve been an Elon fanboy since pre-Falcon 1 times and all the time I could simply follow his first principles to the core, be it the switch to electric cars, flight control using a quorum cluster of non-space-rated processors or the viability of methane and stainless steel for Starship against all previous failures.

With Twitter, I‘ve lost that feeling and I don‘t know how it would be possible: Free speech, happy advertisers, satisfied governments globally, increasing or just persisting audience. There‘s just no way he will keep all those targets at the same time.

It was decried and ridiculed equally at the time, but I doubt the LED revolution would have happened half as fast without the massive artificial demand created by the incandescent bulb ban of the EU. Next Up: Fossils…

True that.

> there's no way of our GDP keeps growing indefinitely while reducing our ecological footprint.

This is at its core not a truth, but really an unsound belief system that could have been phrased the same about water quality, acid rain, leaded gasoline or the ozone hole. Yet these have all been tackled by humanity as a whole already, because we saw a pretty bad impact on our livelyhood quickly.

No doubt, CO2 is really the greatest challenge of all times, compared to nearly any global problem before. But it‘s far from insurmountable, and definitely not directly coupled to growth.

All it would take is a (literal) moonshot effort: Taking the same funds as the Apollo Program (in 2022 adjusted dollars) would be enough for a full transition.

Before this decade is over. Not because it‘s easy, but because it is hard.

Unfortunately, the US seems to have run out of massively visionary presidents since 1968.

Fun technical challenges.

Predictable, steady work schedule.

Nice compensation package.

Pick two (at most). Sometimes you‘re really lucky, but it won‘t stay forever as the organization will drift into one of the corners naturally.

Note it‘s okay to re-pick occasionally as sometimes you need one more than the other in life.

I sure did, and found some peace in this great article from NYT: https://news.ycombinator.com/item?id=15107818

Pretty ingenious, reminds me a bit about how Elite double used its own program code as a lookup table for building a procedurally generated universe in order to conserve memory.

I figured it was ingenious back then but I had no clue it‘s biomimetic :)

https://en.m.wikipedia.org/wiki/Elite_(video_game)

I‘ve noticed this happening not just with characters, but with narratives as well.

Mythbusters used to be my all time favorite TV show for almost a decade. They had such interesting myths (lead balloon!), authentic characters and real builds that also went wrong at times, with some pretty random occurrences.

And then someone from Discovery‘s analytics department figured out they got the best ratings on some of their explosions.

Which lead to this incredibly thought-diverse show jumping the shark by pivoting to basically „let‘s find yet another excuse to blow stuff up“ in the last seasons. Yawn.

I guess it‘s really due to catering to the mainstream. Who said it so well again: A one-size-fits-all solution barely fits anybody.

This. If there would have been any Alpha to gain by dead simple, well known pattern recognition, it would have been exploited very long ago already.

By now Technical Analysis is all about reading meaning into white noise (that’s overlaying fundamentals and a general light upwards trend).

So even if it was this easy and logical, the only logical consequence would be scrapping SLS right now.

Artemis 3 isn‘t just depending on one, but two super heavy launchers that haven‘t even left the ground yet.

And if you put Starship in the critical path, why would you spend billions to introduce another bottleneck that‘s literally just duplicating Starship‘s ground to orbit capability?

If NASA agenda really only was getting the USA to the moon, the only rational choice would be doubling down on Starship all the way.

That‘s exactly what I mean. In their newest plan, they don’t. Someone else does.

Does anyone get the point with Artemis? Because I really don‘t.

To my untrained eye the current outlook on the program looks like that rocket is right now only supposed to make two roundtrips along the moon — and then in that third „Starship HLS“ mission Artemis’ only job is to take up the astronauts to orbit: while Starship (notably designed to be human rated) will be doing all the heavy lifting of the mission.

In this context it all looks like a total sham, it would literally be cheaper and easier to send them up in another reusable Starship.

What other purpose does Artemis have except building the most expensive trip to orbit existing so NASA can say „hey, we did something about this moon mission too!“?

Nice! I used to love AHK to death, incidentally it also got me my first paid programming job.

It‘s about the only software I dearly missed after switching to Mac, but recently I found Hammerspoon and couldn‘t have been happier.

Yep, Lua is a bit weird and exotic, but hey, so is AHK. In the end, you trade in some of AHKs terseness for better modularity and compatibility with existing packages.

https://github.com/Hammerspoon/hammerspoon

As this category used to exist (and partly still does in my country with Miele, but is considered niche), I doubt you‘ll get much product market fit there.

Total Adressable Market is the nerds who care, i.e. barely anybody.

But trust me on the sunscreen!

„Dr. Martin Seligman, past president of the American Psychological Association, believes there are five factors that contribute to our happiness: positive emotion and pleasure, achievement, relationships, engagement, and meaning. Of these, he believes engagement and meaning are the most important. Becoming more engaged in what we do by finding ways to make our life more meaningful is the surest way to finding lasting happiness. When our daily actions fulfill a bigger purpose, the most powerful and enduring happiness can happen.“

— The One Thing

After studying the science of happiness for quite a while, this paragraph resonated a lot with me, summing it all up pretty nicely. Can recommend the whole book!

I‘m pretty sure Matthew 5:38 was written before the advent of corporations, much less the completely deregulated capitalism of 2022 corporate America.

So if you insist applying 2 millenia old wisdoms to your everyday life, consider adding some Matthew 7:12 as well and sue them if they would have sued you in the same situation.

Sounds like still too many techies believe a technical solution is automatically a great business case.

I have observed over the years a tendency of people leaving software for something completely different (often something much more tangible).

Maybe that’s your beef, maybe not.

My take on it is slightly different: It could very well be that instead of burnout, you are just missing purpose and meaning in your life. What is the legacy that you want to build for your kids?

Don’t despair. It’s all good. Give yourself props for mastering all but the highest layer of Maslow‘s pyramid. You’ve achieved something most people on this planet never will.

Onto the last one. Good luck with that!

I think that hits home.

A lot of people would just answer something to the likes of "Well, they made The Matrix with a computer 20 years ago", and technically that's just as true.

From their remote viewpoint on what's happening in IT, the rest is an implementation detail to them.

I think the success really got to him. As a (cured) ex Elon fanboy who used to like him way before he was widely known to a non-nerd community, I distinctively remember this interview around the pre-Falcon 9 times with him which is like

Interviewer: Neil Armstrong says he doesn‘t want a private company like SpaceX to launch humans to tue moon because it‘s unsafe. What do you think about that?

Humble Elon, literally crying: I… I just wish he could come out here and see what we are doing.

That‘s the picture I carried with me for a long time, together with some really smart remarks about climate change on panels that were way ahead of their time.

These days, it‘s really all an ego shitshow, memes and YOLo. He definitely changed a LOT the past few years.

I guess it‘s more telling if people try to use rhetoric devices like trying to force charged, binary answers on a very nuanced, and definitely nonbinary topic.

You like painting people into a corner, do you?

Rhetoric question.

> Mundane chores are all the rage in gaming

Oh, are they? Spontaneous nostalgia ensues…

https://www.c64-wiki.com/wiki/Donald_Duck%27s_Playground (1984)

Why the Fed has made a historic mistake on inflation: Because it didn‘t want to do the other historic mistake.

The Fed‘s basically got binary options here: Keep inflation from ruining lives and savings or saving the economy from collapse. Or trying to do just a bit of both.

I’m rather asking myself: When you‘ve got a water pipe with a valve you can just open and shut and the leakage getting larger and larger with less and less water arriving at the end, is it the one operating the valve doing the historic mistake?

And there we are again: Demonstrating that an absolutist take on free speech tends to destroy free speech itself in the long run.

„The paradox of tolerance states that if a society is tolerant without limit, its ability to be tolerant is eventually seized or destroyed by the intolerant. Karl Popper described it as the seemingly paradoxical idea that in order to maintain a tolerant society, the society must retain the right to be intolerant of intolerance.“

https://en.m.wikipedia.org/wiki/Paradox_of_tolerance

Anecdotally, after two years of literally just looking at tumbleweed in the application inbox, as a still-hiring company I have rarely felt as swamped with really interesting candidates and annoying recruiters as in the past four weeks.

The tide has turned for sure.

School schootings per capita (2009-2018):

USA: 933 * 10^-10

Rest of World: 7 * 10^-10

Having a pretty good statistical significance comparing both buckets, school security and police armament for sure does not seem to have a positive relation with a reduction of shootings.

Nobody wants to hear the solution, but it's pretty clear: If you want to reduce the number of gun fatalities in schools, you need to stop sending people there.

(/s)

And if you feel self-concious about it, just write „freelance projects“ and in the ultra slim case someone still inquires about that telling them point blank you‘d love to divulge it because it was a very interesting project, but you‘re bound by an NDA.