drillsteps5

I very much would like to know how much of this presumably ordered (and backordered) hardware (RAM/SSD/.../wafers) is going to end up being released back to the market when the dust settles. I haven't seen any estimations but in order to put all this hardware to work the hyperscalers need to be building data centers at ludicrous speed. That should be appearing in construction data, jobs data, and many other places. Are we actually seeing any of that? Or is it all just based on the back-of-the-napkin math by Mr Altman and Co and they put all the money they got towards the future projects?

I actually printed this one out to give it a thorough read, something I haven't done in a while :) I'll probably comment on the author's blog later.

My first reaction is that it's a bit of oversimplification, LLMs have been on the scene for what like 4 years now? I doubt the effects will be visible when you look at the data.

But on the topic of inequality as a result of changes how value add is distributed between business owners and the labor - there's a very relevant book by Thomas Pickety called "Capital in the 21st century" (I believe he used the title of a similar work by another economist in 18th century on the same topic). He collected as much data as he could on income and wealth of individuals and groups for Western nations (some going back to like 17th century) and did some analysis.

In a nutshell, the share of profits going towards the business owners (simplifying, inherited capital) in the West had been increasing in the 18th and the 19th century, likely due to Industrial Revolution. Which in the US culminated in extreme inequality personified in Robber Barons (Carnegie, Rockefeller and Co). It was followed by economic and financial collapse, mass unemployment, and arguably, 2 world wars. It also resulted in creating various mechanisms designed to prevent these things in the future, such as anti-trust regulations.

After WWII (especially in the US) the distribution of value add between the ownership/capital and the labor changed so that the labor started getting larger and larger share, to the point that the economists declared that the capitalism solved the issue of inequality. That trend reversed around 1970s/1980s, which coincided with the invention of complex electronic computing and communication devices (and therefore inventions of the new business models). From that point on the share of profits going towards the business owners started increasing again, and that speed has been actually accelerating since the 2000s.

imho at this point US is basically where we were in the end of 19th/early 20th century. The individuals' names are different, the issues (extreme monopolization/concentration of capital) are the same. LLMs and other GenAI stuff are simply part of that trend.

Hopefully people at the power learned something from what happened 100+ years ago. But I kinda sorta doubt it.

Some people strongly prefer to have an external battery that can be swapped (as in pull a tab, remove the battery, and plug in another, which is fully charged). Older Thinkpads (T480 and earlier) had a second, smaller, battery inside that would keep the laptop running while the main (external) battery is being replaced.

I've been using various Thinkpads for 10+ years and have yet to use this feature. But hey, to each his own :)

T480 has 2 RAM slots and can support 64Gb (32+32). The reason it's "unofficial" is that when they designed it (and wrote PSREF) the 32Gb RAM wasn't a thing.

I did a board swap on one of the older Thinkpads. Not that difficult, and the older boards are pretty cheap...

I think it is less of a concern to the businesses buying these things brand new and more of a concern to the tinkerers who buy/repair/resell/use older models. There's a lot of people who still use ThinkPads made in early 2010s (and earlier). I had RAM module fail on an x270 and replacing it only required opening the laptop (RAM sticks just snap into place). If soldered-on RAM fails, it's game over, or at least full board swap.

Plus, no way to put more RAM/replace RAM with larger module if it's soldered on.

Yep. Aiphor's BlueFox NX1 with 4 inch screen is roughly the same size as original iPhone, but has a larger screen (iPhone had bigger bezels and the home button underneath). To me it feels a bit too small for things like typing/texting for example.

Unihertz Jelly Star has 3 inch screen, that's way too small for me.

But they exist and so do people who buy them.

I dived into a niche world of small phones recently while looking for replacement to malfunctioning Pixel 4a (which is apparently now considered compact phone). There's a few small manufacturers in China making some, with 4 inch or 5 inch screen, like Aiphor or Unihertz. And by "small" I mean "they use kickstarter to fund their R&D" small.

Other than that... Nobody's really bothering with compact phones anymore, in the US or in the rest of the world. Bummer.

I switched from Pixel 4a to Unihertz Max (5G phone with 5 inch screen from a small Chinese startup). Love the form factor, I can keep the phone in my front pants pocket again, next to my keys or wallet. I'm somewhat reluctant to put anything sensitive on that phone (like my email), but happy overall.

I'm a simple man, I see Descent, I make sure to mention Overload. Amazing game, likely the first game ever I got to the end just to see how the story ends (yes there is a story and it's pretty good).

I think the Revival studio didn't quite work out, I'm hoping the team is working on something else, they sure do know how to make good games.

Does knowing the system architecture not help you with defining things like happy path vs edge case testing? I guess it's much less applicable for overall system testing, but in "normal" systems you test components separately before you test the whole thing, which is not the case with LLMs.

By "non-deterministic" I meant that it can give you different output for the same input. Ask the same question, get a different answer every time, some of which can be accurate, some... not so much. Especially if you ask the same question in the same dialog (so question is the same but the context is not so the answer will be different).

EDIT: More interestingly, I find an issue, what do I even DO? If it's not related to integrations or your underlying data, the black box just gave nonsensical output. What would I do to resolve it?

>Why did NYC release it in the first place? Did they not QA it?

How do you QA black box non-deterministic system? I'm not being facetious, seriously asking.

EDIT: Formatting

I still install (and use) FAR every time I rebuild Windows :)

Edit: Reinstall, not rebuild obviously.

I have a spreadsheet I've been using since 2017 to track all my spending and savings accounts on a weekly basis, plus some trend analytics, plus some simple graphs on multiple sheets. A few hundred rows and columns, both entered and calculated values (simple formulas, nothing fancy). Haven't noticed any slowness. When I have some data to look at (like .csv or even .xlsx), I always use Calc. I work with Excel at work all the time, it might be faster on larger data sets, but Libre's Calc is more than enough for many use cases.

That didn't work for me. I had my email, documents, some software I used on Windows and files it created (think Photoshop and PSDs). I needed time to find out if I could find similar/equivalent software on Linux, set up new workflows for my work and stuff like that. So I had my primary machine still on Windows and used the secondary to see how all this would work. And then gradually moved all the software, data, and workflows to Linux, until I didn't have anything left that could only work on Windows (well, except games I guess).

As for games, I think a lot of them DO work on Linux but it's easier for me to have dedicates gaming rig on Windows instead of guessing if this new game that I like will work on Linux.

While you _can_ use their launcher, you don't _have_ to. Once you buy the game, you can just download the installer and run it to install on your box. If you want you can save the install package somewhere if you think you'll enjoy the game for years to come and don't want to be dependent on GOG.

I also found out that they have quite a few fairly recent games. Maybe not the top-10 big budget (however they partner with RedProject so they do have Cyberpunk) but they have plenty of solid indy games from 2015 - 2020, and some more recent.

Give me back the proper search which last worked Windows 7-ish (or XP?). And traditional Office interface (not stupid "context-specific" tiles).

I think Linux is not quite there as gaming system. Simply due to games' compatibility (and I don't play latest and hottest titles, more like Cities-Skylines/Transport Fever/Anno/Satisfactory etc). Plus to my knowledge NVidia drivers are still an issue.

But for literally anything else I think it's ready. Just browsing? Office work (writing/spreadsheets/presentations/email)? Development? Media production? You're good.

For Linux-curious I'd advise to get a dedicated hardware, like 5/7 year old business machine (Thinkpad or even smth like Dell Latitude), they'll be under $300. Don't do Arch (unless you do that for the sake of being able to install Arch). Instead, get Ubuntu, Mint, Fedora, Debian, Zorin (the last one specifically for Windows users), or one of many other beginner-friendly distros, and drive it for a bit. Get the software you want, see if it works for you, and if you don't like it, it's all good. If you do, you can gradually move all your stuff to the new machine, or install Linux on your main machine.

That's what I did (quite a) few years ago when I got fed up with Windows 8, took me about a year, but I've been on Mint Mate ever since. My gaming rig is still Windows 11 but all it has is my Steam collection.

Makes sense.

The counterpoint would be that when they started to build LLMs they must have clearly seen limitations of the approach and proceeded regardless, and achieved quite a bit. So the approach to introduce continuous (in-vivo if you will) self-guided training AND multiple sensors and actuators would still be limited but might yield some interesting results nevertheless.

ANN is an attempted model/ripoff (turned out to be extremely simplified but still) of a brain, why not go further? Continuous autonomous learning (which requires continuous feedback in a way of good/bad stimuli) is clearly what makes it work.

The current approach of guided pre-training and inference on essentially a "dead brain" clearly causes limitations.

Someone should just build an ANN as big as currently as possible with current hardware, while still having both inference and training to be as close to real-time as possible (micro-to milli-seconds), build the self-learning using some loose equivalents of pain/pleasure feedback in actual brains, plug sensors and actuators from some sort of robot, and just see what happens.

I think anything less than that is just a parlor trick.

If it's 5 person company they likely don't have HR or recruiting and the CEO is likely doing the hiring (for VPs/Directors/etc). In that case of course you would communicate with them directly, they are effectively a hiring manager and don't have HR to outsource the hiring to.

If the company has a person/group dedicated to hiring then going around them is counterproductive. IMHO of course!

Soooo... My guess is the next step in evolution of automated resume submission apps will be looking up hiring manager (or above) and sending a tailored email to "decrease false rejection rate". Can we expect a "Show HN" post with this feature soon?

You got me there, I don't have one. I'm senior in my career but haven't been able to build a network that I can rely on when I'm looking for my next role, so I'm in the same boat with all of you guys and gals.

I do maintain though that cold reach out is more often harmful due to the barrier of HR/recruiting built to prevent this from happening, and you trying to go around that will likely cause trouble.

Also, when you reach out to the employees for "the referral", what does this even mean? If the person knows you, worked with you, or went to college with you, then they can refer you, but if they haven't even met you and don't even know if you're real, what are you asking them to do? "Hey boss I got this email from this guy he says he's a good fit for this role we have open, do you wanna hire him?", is that it?

DO research the company. DO research the role, the team, the manager, the environment, the toolsets, the issues they're facing. Do NOT flood people's inboxes asking for "referrals" whatever that means.

I honestly don't know if it was sarcasm or if you were serious.

At any rate, the downside of this is as follows.

The goal of having Human Resources, talent acquisition, recruiters, and other similar roles is to let hiring managers (and everyone above them, up and including the CEO) concentrate on doing their job and only assist the aforementioned roles in hiring. Of course hiring manager is ultimately responsible for hiring a good candidate but they are not expected to do things like posting job descriptions, initial screening, background checks, referral checks, employment history verification, dealing with legal stuff like NDAs etc etc, that's the job of HR/recruiters. Candidates reaching out to hiring managers (and especially higher ups) are not treated nicely by the HR as these candidates are attempting to take HR out of the picture.

HR are people and want to keep their job and get paid, and you circumventing them might be perceived as a threat to that.

That IN ADDITION to a disruption you will be causing hiring manager (or especially CEO) cause now they need to decide what to do with your email. Even though they have HR/recruiters to handle these things.

A typical result of such a "reach out" will likely be forwarding this email to HR and subsequent rejecting/blacklisting the candidate.

Edit: some clarifications

The negative effects of social media on teenagers have been researched and are well documented. I have 2 teenagers and honestly don't even need that, I see it with my own eyes.

Ditto. Social Media investigated negative effects of social media and found none. :)

> Instead of applying broadly, identify 5-10 specific opportunities you genuinely want.

Do that.

>Get in contact with current employees at the company. It is important that you send more than one email.

Don't do that.

>I've gotten dozens of emails asking for meetings and referrals.

I've never gotten one in my entire career, and I was hiring manager in multiple companies/roles.

>If the company is <30 people, reach out to the CEO directly.

Don't never ever EVER do that.

Edit: formatting

The primary _stated_ purpose of the technology is to eliminate as many jobs as possible to increase profits for the business owners.

Which is a valid purpose, if I were a business owner that's what I would have done, too, but I wouldn't expect people whose jobs and income are at risk (which is a large part of society) to cheer for it.

What a wonderful way to start my Sunday, to distract from the antics of the rich and powerful, and to pay attention to things that matter.

I found it curious that a novel published in 1989 refers to the future immortal being going through crude pre-singularity historical records and comparing them to cobweb ("An analogy to cobwebs made her smile for a moment."). I thought Berners-Lee wouldn't launch his first "World Wide Web server" for another 2 years, no?