time_to_smile

> These cards are accepted by society because they benefit those who know how to use them

No, they're accepted because the only way we can figure out how to continue economic growth is through a massive credit system. This means doubling down on a strategy of infinite growth as that is the only way such a system is sustainable.

Cheap money coupled with expanding credit is the foundation of the global economy for the last decade. You could also argue pretty easily that cheap money is just another view of expanding credit.

If we increased the cost of money and reduced the amount of consumer credit available the global economy would collapse. Ever since the GFC we've pushed off collapse only by expanding global debt.

> they can spend it, or they can save it.

The trouble is they haven't been saving it. Savings rates is near the lowest it's been in years [0], debt is rapidly increasing [1] and we're seeing quickly increasing dipping into 401k savings [2].

I do agree with your general argument though, that this isn't about workers being paid to much. All evidence suggest that despite the increases in worker pay, more and more American's are stretching themselves to continue consuming and we've seen in plenty of places that consumption is dropping.

0. https://tradingeconomics.com/united-states/personal-savings

1. https://www.cnbc.com/2022/11/15/household-debt-soars-at-fast...

2. https://www.wsj.com/articles/short-on-cash-more-americans-ta...

We've worked to keep the dollar less strong since a rapid increasing in the value of the dollar risks a global debt crisis. A lot of people in finance were quite concerned earlier this fall when the dollar started to rise rapidly.

I regularly keep up with financial news and it's hilarious how each morning financial reporters quickly rush together articles about "<direction market moved> caused by <current top news story>".

I really enjoy the pre-market opening reports, always trying to predict the direction of the market (usually based on the current futures direction) that are, as expected, wrong about half the time.

The best are when you can catch two contradictory headlines based on when they were published relative to market movements.

A stock movement that's within two standard deviations of typical price movements is very likely not the result of a single new piece of information.

That's why the time to make demands and organize was when this current tech boom was just getting started, when engineer salaries started to really pick up and before bootcamps and massive new waves of CS grads where in swing.

By organizing across the field, software engineers would have been able to have more input into the rate of engineer growth (which would have hurt bootcamps but generally lead to higher quality engineers, and avoided over hiring), compensation (which likely would have hurt some of the people making insane FAANG salaries, but lifted up most others), professional licensing (i.e. maybe you just do that leetcode/project/etc gauntlet once to get licensed, not for every new application), enforced professional ethics (as a profession we agree to limits on the kinds of exploitative work we do) etc, etc.

The problem is that anyone even suggesting this back then (and there were a few people) would be downvoted into oblivion. Engineers found rewarding, high paying, prestigious work easily and virtually everyone thought that was an eternal truth.

Now it's way, way too late.

He's not wrong. The key ingredient in ChatGPT is not brilliance but capital. It takes a lot of money not just to do the raw training but to get all the data together and process everything.

There's no break through insights that make ChatGPT work, just a lot of consolidated wealth.

The hacker part of me finds AI less and less interesting for this reason. We're seeing what the limits of pouring resources into the problem are. The results are cool, but I think we'll very soon see progress bound since we're using most of the data we can and it doesn't look like adding even more parameters to these models might not yield that much of result.

But this shift in AI means it's increasing something programmers consume rather than produce.

As far as interesting AI goes, I found the series of posts on stuff people were doing with Prolog that showed up during the last month much more interesting.

This is a weird take. Are you arguing that we aren't seeing an abnormally high number of layoffs in tech? That this is just "media sensationalism"?

Nearly all the layoff reports I've come across I hear about before they're in the mainstream media. Additionally I predicted months ago that we'd be seeing massive layoffs in tech, before the media was talking about it at all.

These layoffs are real, are atypical for what we've seen the past decade, and are impacting a large number of real devs.

Honestly I haven't even seen the media talking about it until the last few weeks. When I talked about impending tech layoffs back in Oct-Nov nobody outside of the tech community had any idea that there were any issues facing tech.

I hardly see how layoffs are being "played up" rather than just being reported.

Chewy hasn't solved the problem that Pets.com had, we've just had enough free cash flowing around for longer this time so it doesn't matter.

When Chewy starts to make more than they spend then it will be interesting, but as it stands now it just looks like we're in a bigger bubble rather than solving problems that couldn't be solved back then.

Startups tend to be shockingly ignorant of the culture at larger companies (or non-startups in general), and personally I find it a bit of a red flag when it seems like nobody at a startup has worked for a real business before. It's great to have fresh takes, but there should be a few people on the team who also know how normal companies operate.

I remember interviewing for a startup that sounded pretty promising. The interview was going well and then I was asked for a presentation. I told them that I was more than happy to present on personal projects (of which I had plenty) but was not comfortable sharing details of work related projects while not explicitly representing that company.

They basically said "too bad" and where shocked that I cancelled the rest of the interview. I've worked at plenty of a companies in the past where being willing to share other companies' internal work would be the red flag, and turning down such a request would be seen as a good sign.

> I would actually not apply to this company.

Sadly that was the same impression I've had, and this is from someone who has been following the author's writing for awhile. Especially in the current economic environment.

> These articles claim that at large companies that receive an influx of resumes, recruiters set rules to surface resumes that contain certain keywords... That is a bad strategy for applying at startups like ours. We’re not looking for keywords. We look for demonstrated expertise. Here are a couple of ways to demonstrate your expertise.

I'm always a bit surprised when startups, whose leadership doesn't even seem to understand the dominant hiring patterns for large, mainstream corporations, expect candidates to cater specifically to them.

If you're trying to get me to work for an "AI/ML platform" startup in this market right now, you'd better spend more time convincing me that you'll still exist in 2 years than telling me how I can tweak my resume for this specific company.

Honestly, a startup looking to hire founding positions should probably be searching their own networks rather than screening hundreds of resumes a month.

> and now rarely eat out.

That's the inevitable outcome of this type of behavior. Like so many startup's attempts to squeeze customers in the short term, it works for a small window of time but leads to longer term consequences that are never properly accounted for.

For most tech workers an extra 20% on top of everything else is not that big of a deal, but for anyone who saves up for a night out that can be a big upset.

The only companies that can do this drip pricing and hope to get away with it long term are essential services without serious competition (for example telcos). All of the rest of these drip pricing schemes will come back to hurt companies in the long run.

Has everyone else been using a different tool than me? ChatGPT is interesting, but is laughably bad at any non-trivial instruction.

It's fantastic at generating content that on first glance looks remarkably right, but always fails fine inspection.

All I've seen from LLMs is much better demos of the types of funny things markov chains were generating two decades ago (anyone remember the various academic paper generators?). However I have yet to see anything that stands out as really remarkable.

My read is that people want to see incredible progress towards strong AI, and LLMs do a great job of letting people feel like they're seeing that if they want to.

I suspect in 5 years we'll largely have forgotten about LLMs and in 10 they'll come back into popularity because techniques will become more efficient and computing power will increase enough that people can train them on their home PCs... but they'll still just be a fun novelty.

If you want a more purely academic/economic perspective on the larger issue I recommend "The End of the Everything Bubble" by Alasdair Nairn. As implied by the title, it doesn't focus exclusively on tech but on the asset bubble as a whole.

I'm surprised people working in tech can not see this coming, haven't you ever looked at the current tech industry and thought to yourself "something is not right here?"

Tech has been a massive game of shuffling around VC money to create the illusion of a massive part of the economy when it's really all the same money being shuffled around looking for profit and failing to find it.

The heart of the problem boils down to this: There are basically two types of tech companies a.) unprofitable ones, and b.) profitable ones who make most of their money from companies in group a. For over a decade investors have been focusing more and more on growth while completely forgetting the basics about how a business need to run.

This same issue happened in the dotcom bust, PG even has a great description of this from a earlier essay [0]:

> It was not just our price to earnings ratio that was bogus. Half our earnings were too. Not in the Enron way, of course. The finance guys seemed scrupulous about reporting earnings. What made our earnings bogus was that Yahoo was, in effect, the center of a Ponzi scheme. Investors looked at Yahoo's earnings and said to themselves, here is proof that Internet companies can make money. So they invested in new startups that promised to be the next Yahoo. And as soon as these startups got the money, what did they do with it? Buy millions of dollars worth of advertising on Yahoo to promote their brand. Result: a capital investment in a startup this quarter shows up as Yahoo earnings next quarter—stimulating another round of investments in startups.

I remember the dotcom bust and the big difference between then and now is that the duration of the bubble was much smaller during the dotcom. People back then proclaimed we where in a "new economy" where profits didn't matter anymore. The boom was only a few years and so those critics where proven correct in a short window of time. This time we've been in a bubble so long people don't even remember that companies need to make profits and be sustainable.

During the period of cheap money VCs flooded the industry with so much cash it started to get weird. Wayfair is a great example of this, they had 150 data scientists at one point, what in the world does a furniture reseller need with that many DS people?

Now we're starting to see the conditions that allowed this bubble to form relax very quickly. Not only is cheap money no longer a thing, but the consumers who have been gleefully pouring their extra money into apps that don't quite make sense are starting to feel their wallets get tighter (actually the data looks like they're starting to max out their credit and being forced to reduce spending).

And that's just tech. We truly are in an "everything bubble" [1] were asset classes across the board are wildly overvalued. Unlike the dotcom, this will not stop with tech.

0. http://paulgraham.com/bubble.html

1. https://en.wikipedia.org/wiki/Everything_bubble

It's not doomscrolling, its reality. The writing has been on the wall for over a year, and yet people still refuse to see it.

We're at the beginning of a major contraction in tech. It might not be pleasant to hear about but if you work in tech it's important to pay attention.

It really saddens me how often the HN communities general reaction to unpleasant news is to immediately stick their heads in the sand and claim everything is fine.

I find one of the biggest mistakes programmers have about mathematical notation is that it's somehow just a terse, badly implemented programming language. But this is a very poor understanding of what mathematical notation is doing.

I think this error in thinking comes from the fact that Sigma notation can often be trivially implemented as a for loop.

Programming languages are designed to describe a specific computation, whereas mathematical notation is typically trying to describe an idea (one that might not even have a implementation!) Notation only sometimes and coincidentally describes computation as well.

The ambiguity, implied variables etc are an essential part of mathematical notation in the same way it is in common spoken language. Mathematical notation exists to help abstract and work out very hairy ideas, and often that ambiguity is necessary to show connections.

> code should be optimized for readability, not writtability

Mathematical notation is readable if you're literate in it. It takes lots of practice to become fluent in it, but once you become more familiar it's much easier to read than text (which is why it's used in the first place). Mathematical notation is an extension of mathematical writing, not computational implementation.

Reading mathematical notation is much closer to reading poetry than reading code.

I highly recommend working through Claude Shannon's Mathematical Theory of Communications [0]. It's originally a paper but was later restructured as a book, in either form it works quite well.

The reason I recommend it is because it shows mathematical reasoning that is easy to follow and relevant to your daily life. It's real math, but very easy to read through and understand. If your unfamiliar this paper is where the very idea of "bits" comes from.

One of the most important things in the paper for non-mathematicians to see is that the definition Information Entropy is derived simply from the mathematical properties Shannon desires it to have.

This is important because I find that one of the biggest questions people ask about mathematical formula and idea is "What does this mean? Why is it this way?" without realizing that math is really not engineering nor physics. When deriving his definition of Information, Shannon simply states that information should have the following x,y... properties and then goes on to show that the now standard definition of information meets all these criteria.

In mathematics it is very often the case that only after an idea is created to we start realizing the applications. This is quite different than science where a model is only adopted if it correctly describes a physical process.

Work through the paper and you will have worked through the mathematical underpinnings of the information age and will likely have understood most of it pretty well.

0. https://people.math.harvard.edu/~ctm/home/text/others/shanno...

I think year here is also super important. In 2008, even without the benefit of hindsight, investing in tech is not a crazy idea. We had a real-estate/banking crisis and there was a lot of people looking to figure out where to put their money.

Tech in 2008 looked very healthy, and VC funding only went up from there.

We're looking at an actual contraction in the tech sector now. We can argue how about whether this is a small "correction" or something more major, but I think it is a good idea to not just blindly apply good market advice from 2008 to the 2023 landscape.

It probably is a great time for a startup, just probably not a tech startup.

I think you're half right, but missing a big piece of the puzzle.

I agree with you that people's spending habits don't make sense. I see young people complaining about having to live with their parents when near their 30s who still think a two week vacation with a nice rental and eating out ever night is a basic human right. I'm constantly horrified how many people simply cannot cook their own food. The wouldn't know how to make a good rice and beans dish if they had to.

But the part you're missing is that our current system cannot survive without this spending. That FAANG job that paid you so well, where do you think all that cash came from? My friends making money for insta are basically making that money selling avocado toast.

If people started living like their great depression era relatives they wouldn't be able to afford houses because our entire economic system would break down.

> We are nothing like that situation

Agreed because this time the asset bubble extends well beyond tech. This is a much larger problem and will take much longer to unwind, cutting much deeper than that.

I like the optimism but I suspect we're still in the early stages of this process. There's lots of bubble to unwind here, and this mess hasn't even started to untangle.

I recently came across this quote from PG in one of his earlier essays [0] describing the early days of the dotcom bust regarding yahoo. Sounds very similar to the situation we're in now:

> It was not just our price to earnings ratio that was bogus. Half our earnings were too. Not in the Enron way, of course. The finance guys seemed scrupulous about reporting earnings. What made our earnings bogus was that Yahoo was, in effect, the center of a Ponzi scheme. Investors looked at Yahoo's earnings and said to themselves, here is proof that Internet companies can make money. So they invested in new startups that promised to be the next Yahoo. And as soon as these startups got the money, what did they do with it? Buy millions of dollars worth of advertising on Yahoo to promote their brand. Result: a capital investment in a startup this quarter shows up as Yahoo earnings next quarter—stimulating another round of investments in startups.

0. http://paulgraham.com/bubble.html

As someone who entered the work force in the shadow of the dotcom bust, I think the biggest piece of advice is that people who have only lived during the time of the "Great Bubble" don't have advice much to give you.

Entering the work force now or in the next few years, a new grad will very likely have an easier time navigating the job market than people who have spent nearly a decade working for bubble-frenzied, VC backed companies.

In the years following the dotcom bust I new tons of software folks desperately trying to reconfigure their careers. It's not hard for a fresh grad to change course, most people don't end up working in a career that is directly related to their major.

Entering the work force will be harder and harder but new grads will have less expectations and bad habits they have to overcome to find an occupation that works for them (they also won't have no-longer-reasonable comp expectations).

> totally worth keeping 5-10 person engineering team

I've worked in AI/ML for a long time, and I can tell you that in all but the rarest of exceptions nobody has sat down and done the math to really determine if that value is there. In my experience, most cases it's not.

I've seen teams like you describe and then asked them "what's the difference in value between what our model does today and the theoretical optimum"? The answer, which nobody liked, was fractions of a penny per user. That means if the team achieved perfection it wouldn't really matter in practice.

We're going to see a lot of the AI/ML teams disappear as companies are forced to focus on determining the real value add being provided.

Wildly overpriced compared to building it yourself or compared to other PC building companies?

I purchased a Thelio Mira awhile back and have been happier with it than any other PC/laptop I've owned before.

The big value add that System76 is having a Linux desktop that is guaranteed to just work out of the box the same way you would expect and Apple machine to work out of the box. In fact, I've probably had less issues with this machine out of the box than any new macbooks I've had in the past few years.

I've built plenty of my own PCs before and run linux as my primary desktop multiple times, and my Thelio is a wildly better experience than in the past. The build quality is excellent and for the first time I can really use Linux as my primary desktop with no problems.

Building a PC yourself is always going to be cheaper, and for many people that's the preferred path anyway. The point I'm at in my life I would much rather pay a premium to not have to worry about that at all (especially when it comes to hardware on linux).

If anyone is look for a "just works" linux PC, I've found system76 to be a great experience.

Before answering your question, you need to first be clear what "sustainable" means. For me sustainable means that we can continue to life as we do indefinitely.

Our current economic structure, due to its reliance on credit, requires perpetual growth and development in order to pay off today's debts. Debt in all forms has been growing increasingly and rapidly in recent years.

Infinite growth is not possible on a finite planet.

In many areas we are already seeing the limits of growth, from strains on oil supplies to global population growth starting to slow down. This is already, today, putting a strain on our economic systems.

Since our current way of life can only be sustained by future growth, it is by definition not sustainable unless you sincerely believe growth to be without limit (this would require near term interplanetary travel and energy advances such as fusion). As mentioned, we are already seeing system strain suggesting we are hitting limits.

Contraction is the preferable path of the two realistic alternatives, the other is complete collapse.

How do you propose the economy growth for the foreseeable future and becoming more sustainable?

Or we see a contraction in globalization in general in which all economies shrink.

It's entirely reasonable that we could enter a period of long, slow decline across the board. Especially as we continue to push the limits of natural resources and global supply chains.

For example suppose the US continues to move its push to return chip manufacturing to the US. This might mean both that US chip manufactures have a more healthy future than other more fragile tech companies and that they shrink in size. We could see a return of manufacturing to the US which leads to continued employment in US labor for while also meaning that labor force gets paid much less.

We're already starting to see evidence of this happening.

The concerning thing is that I'm not at all sure that our incredibly debt dependent global economy, which assumes future growth, can really handle a gradual contraction to a more sustainable economic structure.

Either way, assuming up is the only way for the market to go is a very naive assumption, but one nobody is happy questioning.

Do engineers not take calculus anymore these days?

Gradient descent uses the gradient, but is itself not the gradient.

The gradient itself is just the vector of partial derivatives with respect to each parameter.

Gradient descent is just following the gradient "downwards" towards a minima.

The article even describes it pretty well in non-mathematical terms.

> the gradient of the distribution (think of it as the slope of the high-dimensional surface).

I always hear devs say they never use calculus but I suspect it's because you can't use what you don't understand.

> I don't know how many "that's what was said of Airpods" are needed for this argument to die

Apple has always been the exception that proves the rule in the types of cases. Very, very few companies are able to create demand rather than merely satisfying it.

On top of that only experience with Airpods is as the current gen teenage reality blockers and older people screaming through static about how they must have run theirs through the wash.

> snag a Boosted Board

Boosted went out of business in 2020.

It's funny how many times I've seen a "revolutionary" transport product released in the tech word with the exact same set up and experience: I see a hyped up marketing video from the company, then look on for the inevitable wired video article trying the product out (that is clearly also mostly PR). All without seeing the price.

Each time I think "hmm... I'd spend 250-500 to try that out if people like it" only to find out that the true price is close to an order of magnitude more than what my initial estimate.

It was the same with Boosted as it is with these and the many other similar products I've completely forgotten about.

I really hope that I get laid off in the first round of these layoffs where everyone gets good severance.

My current worry is that the first round of layoff at my first company will get 4 months severance, I'll miss that round, and 2 months later get laid off with only 2 months severance. This would of course mean that my severance ends at the same times as the first group but I had to work an additional 2 months.