switchbak

What I'm getting from all this is that we have an armchair expert explaining details about something they're not an expert in, and we have you reviewing the way they present this information, saying their very weak hedge wasn't strongly negative enough. Based on priors about the industry.

Basically, HN doing HN things - having a good deep think about topics outside their realm of expertise, then debating the minutiae. I'd just rather trust someone who has spent their career researching this stuff, thanks (unless that's what you do - in which case please tell us more).

It was certainly the case that gym bros thought you had to cycle it. But you're right that this was never necessary.

"poses no danger to your hair" I keep hearing folks say this, and even in this thread there's someone who's experienced that and had it stop when they eliminated creatine. I too have experienced exactly that with exactly the stopping of issues once I stopped taking creatine.

Perhaps you can say "danger to your hair is rare", but there's very clearly (to me at least) an effect there, and the way it gets hand-waved away is strange to me.

Human ingenuity thrives on constraints.

If I worked at a place like that, I'd sure as hell work my butt of to get a job somewhere else.

Or in my case, actively ignore any and all recruiting from that sesspool.

"that can grammar correct" ... Did you do that on purpose?

For the near future it seems that the new models will consume whatever improved hardware capacity we have. Competing with that is challenging, but I also think there will be strong economic incentives towards cheaper but adequate models on other providers.

I don't think we'll see home users being able to match even the low end clouds for a long time.

Longer term I think we'll see these uses of AI cluster into a few groups:

- maximal code / reasoning quality, at high prices (Fable)

- typical code / agents (sub-Opus, Terra)

- cheap but decent enough quality (think Deepseek / GLM / Luna)

- so cheap I don't care about utilization (Deepseek, and friends)

And also more niche ones:

- ultra fast with high quality answers (typically sub-SOTA). Cerebras / dedicated silicon type approaches, expensive.

- ultra fast with mostly-adequate answers, and an openness to retries, moving up to better models

I think the open models will dominate (not with individuals, but low cost providers) all except the top 1-2 of those categories, and there will be a continuous erosion on the big player's moats. The top categories are also where all the money is, but I'm not sure it can justify those investments long-term. I also think they will have to squeeze more money out of them to justify the investments, which will also drive people down the list.

Edit: clarifications.

Supposedly it’s an urban (rural) legend going around for a long time. Good story, even if it’s not true.

The prison sentence for being caught doing legal things is surprisingly short.

It is when folks like you come around and vomit this kind of thing.

"while others have their own battles, they are very rarely similar to this" - meh, I don't think that's true. Being human is hard, and in some circumstances can be very fucking hard. If you've had a challenging time in your youth, that makes it exponentially harder.

That's most people. They just keep it inside, assuming everyone else is doing better than they are.

The mass of men lead lives of quiet desperation.

The Buddhists call that seeking ground, and they consider it to be a real source of suffering. Their take is that the reality is fundamentally groundlessness, and the more you can relax into that (easier said than done), the less you'll suffer when the inevitable shit hits the fan.

As someone who has had a metric ton of shit hit his fan, this is a hard pill to swallow, but very effective medicine and has helped me tremendously.

In modern terms, you could say it's good to embrace a growth/flexibility mindset and work on the things that help you to build and restore resilience.

Good refinement.

I like Paul Graham's "keep your identity small" piece: https://www.paulgraham.com/identity.html

When you're not personally identified with your work, it stings a whole lot less when someone rips it up. And keep in mind we're imperfect people working with other (often very) imperfect people. It's a mess, and always will be.

"HN comments are _NOT_ going to help debugging your mental state" ... I mean in aggregate, sure. But of course there could be people that are offering useful advice, in a well meaning way - that could actually help OP. You're being really negative here, and I think it's unwarranted.

Sounds like it grew out of the rationalist community (eww), and got picked up by various podcasters as it became a phase du jour.

See https://www.yuzeh.com/etc/2021-01-24-beliefs-as-structures - folks have been using it for a while now.

Now everyone is saying "hits different", and that one drives me crazy. Old man problems.

Thank God, we can't just have people getting high for free.

Here in Canada you're not allowed to 3d print a modded grip for the rifle you're not allowed to buy, but in the USA you can get OPIOIDS at a GAS STATION! ... things are definitely different down there.

Also you can just walk in a farmer's field, pick a mushroom out of a cow patty and boom - you're high AF! Someone should regulate cow shit, there's little to no control over it!

There's lots of evidence for positive influences of Caffeine on the body (some negative as well), especially the brain. In particular, there's active and promising research on the neuroprotective effects for Alzheimer's [1] and Parkinson's [2].

"There is a wealth of accumulating biological, epidemiological and clinical evidence to support the further investigation of selective adenosine A2A antagonists, as well as caffeine, as promising candidate therapeutics to fill the unmet need for disease modification of PD."

1: https://www.mdpi.com/2218-273X/13/6/967 2: https://pubmed.ncbi.nlm.nih.gov/33349580/

"The point isn't whether it's true about you in particular" - then wtf are you saying? That this person can say true things but is still wrong?

I agree with them, and I also agree with the concern in the article. Both can be right. And I think you're off base here with all your unfounded assumptions and straw-manning.

"Load bearing beliefs" is a thing in the podcast/YouTube world at a minimum. Perhaps you're not as online as some other folks (probably a good thing!).

I think when it comes to watching the night sky, weighing to populated areas isn't necessarily right. It's not like you should value rural folks' view of the night sky less.

Paraphrasing (strawmanning?) the original - it doesn't matter if StarLink has ruined the night sky if city dwellers can't see the it. Well, I believe it still matters a lot. And we're still talking about nearly 1/2 the world population here.

“Remote areas” make up most of the world.

That's mostly a value judgement though.

What about fiction vs non-fiction? Travel vs escapism? This is hugely subjective, and very personal. What one person finds educational could be a navel gazing time wasting pursuit to someone else.

"the "deep" thinking associated with reading has always been a bit of an elitist idea" - precisely. This is all about massaging the categories such that consuming or producing a given type of media in a particular style is "good" and everything else is inadequate.

So if I read a bunch of tech manuals, I'm not a reader because it's not fine literature?

What if I read them as PDFs? What if I print them? Where's the line?

I think we ought to call it something other than simply "reading", because the author seems to be leveraging the dual-meaning of that word to make their point more strongly. But "consuming literature for enjoyment" doesn't come off quite as spicy as pretending that others that don't are illiterate.

Java's RMI sounds similar ... but I haven't really seen much of that used in a robust way. You can only ship serializable state which becomes quite limiting after a while. Shipping the data and reusing the same code across nodes seems to accomplish the same thing with less headaches.

You need to ramp up, of course. Given you're training at an appropriate rate, and you're doing sensible things - you should be fine. By sensible things I mean: short and intense weight training, intense cardio sessions, and not-too-long endurance sessions (eg: < 5h). More endurance is fine, but it can be very fatiguing, especially for the older athlete. If you ramp up to those over time, you needn't worry about "free radicals" (that's an oversimplification).

Recent research highlights a number of things, but what sticks out is that it's important to maintaining muscle mass because we lose a bunch as we get older, and loss of muscle is a leading morbidity factor.

They did say “legal supplements”, not prescriptions. But I’m not sure what they’re actually suggesting.

There are (illegal) substances that are highly effective for treating these issues, PTSD, etc. Psilocybin, ibogain, etc. Obviously when administered correctly, with therapy, etc.

It seems to me that cannabis users aren’t seeing the benefits of the aforementioned group. My experience of cannabis stoners is that it’s used to numb out and for escapism, which certainly aligns with what you’re saying.

That’s valuable: I thought the whole thing was dismissed based on the brown dwarf explanation, this is interesting that that’s not the case.

It’s hard to keep up with all the news, particularly so given all the slop floating around. YouTube in particular seems to have a bunch of AI channels that just talk about this endlessly.