Consolidation of a vibrant industry to just a handful of major corporations is a real loss, in my opinion. It's depressing in part because it makes avoiding those corporations much more difficult or impossible, because it makes the industry less vibrant and responsive, and because it locks a lot of talented people out.
It's also more than a little dystopian as the corporate stranglehold over all of us tightens just a bit more.
I don't think that's exactly what the article is stating. It'd be really hard for an upstart to build a formidable competitor to AWS for example, but that just means that the next breakthroughs will be less about infrastructure and more about applications (not apps, but different use cases). "Innovation, uh, finds a way."
Perhaps I was misled by the article analogizing to the auto industry. The auto industry is a great example of exactly what I fear is the future of of the software industry.
I'm not sure sure the auto analogy is very good, either.
Today there is the choice of a large variety of safe, efficient and performant car models from insanely low prices (once inflation is taken into account), and soon there will be a similar range of choice in EVs.
Generally speaking, quality is vastly higher than it has been in most lifetimes too - when did you last see a new car with rust on it, for example? (A few brands, notably Mercedes Benz, produce far worse products than they did in the 90s, but they are the exception).
If a wide choice of quality products is the future of the software industry, I'd welcome that, since it certainly is not the state of it today.
I am certainly not saying there aren't some advantages to an oligopoly, as the auto industry illustrates. But I do think the disadvantages exceed the advantages by a very large margin, even in the auto industry.
For instance, there are aspects of new cars that are ubiquitous and I find so objectionable that they prevent me from buying one, probably ever. If the market were more vibrant, I suspect that I'd have real options aside from "I can't have a newer car".
Also, oligopolies don't automatically mean that the products are improved. Usually, it's just the opposite.
The auto industry isn’t really an oligopoly though. Even if you restrict to “manufactured in the US” there are more options now than there were 30 years ago _because_ the three big manufactures were making trash, and competition did the job it was supposed to.
> there are more options now than there were 30 years ago
I suppose how true this is depends on what sorts of options you're looking for. The number of acceptable cars on the market now, according to my criteria, is lower than it has ever been.
This. And VC doesn’t help much, because as soon as you take the money you are in a race to become the next overpowered, uninspiring behemoth, only with less resources and a low chance of a real payoff.
Consolidation may not be inevitable, but it did happen on so many previous innovation waves (steam, electricity, telegraph, telephone, automobile, aviation beginnings were full of small, growing, competitive ideas and ventures too).
It comes with industrialisation and commodification and norms and regulation, so it's not all bad either as it consolidates/improves the general quality and availability.
Well, if you want to take the traditionally Marxist position, all that competition results in duplicate work. During the golden age of the workstation, Sun, Silicon Graphics, DEC, HP, NeXT and IBM all made workstations with separate, incompatible operating systems and CPUs. In the end, all those vendors died, and that effort was entirely wasted. It would have been much more efficient if the government had designated a single workstation vendor from the beginning, and denied CPU purchasing licenses to anyone who wanted to compete with them.
Efficient? Maybe. But without competition of any kind, there's a possibility we'd still be using DOS lol. There can still be advancements of course under such a system, but I figure the progress would be a lot slower.
On the flip side, I can almost see part of your argument. The weather is a perfect example of your argument. The US government funds all the weather models out there, but due to lobbying, they're not allowed to actually have a website that tells us the weather. Instead we get like 5 crappy ad-infested sites that are terrible to use and just make a few people rich. It's insane.
> It would have been much more efficient if the government had designated a single workstation vendor from the beginning, and denied CPU purchasing licenses to anyone who wanted to compete with them.
How would it be more efficient without a way to know if the end result is any good?
The problem with comparing to car companies is that there are much tighter limits on how much accumulated power can corrupt a car company.
Sure, they got too big for their britches and created propaganda campaigns against pedestrians. They set up a status quo re: the environment that future generations will hate us for. But then their ability to be evil hit a plateau. The cancer that ought to have killed them dwindled, and for better or worse, we collectively said "this is fine". As the head of a car company, your paths to god-king-of-the-universe are significantly limited at the top end, and therefore so too is the extent to which your power turns you into an ineffective lump.
Information companies are not so limited. So the cancer that comes with being powerful is much more likely to be terminal. That is, there's nothing stopping Google from driving their user base away by just showing 100% ads and 0% content. They'll get worse and worse until we're all throwing money at anybody who stands a chance at putting them down.
Just look at Elon Musk, for whom efficacy appears to be secondary to simplify being the biggest threat to the status quo. That dynamic isn't going away: When he ceases being The One another anomaly will emerge.
So sure, it'll get worse before it gets better, but it's not going to be permanently worse. Even if we don't shoot them first, the bad guys will shoot themselves in the foot eventually.
> So sure, it'll get worse before it gets better, but it's not going to be permanently worse. Even if we don't shoot them first, the bad guys will shoot themselves in the foot eventually.
I used to think this way, but I have to admit the last decade or two has made me much more pessimistic about this.
There are just too many examples of oligopolies that have never stopped being terrible. I think the auto industry is one of them, and I suspect the software industry is likely to be another one.
I've been thinking this for a while now. But I don't think that means the disruption is done. Think about how TikTok has disrupted Facebook or how Salesforce has disrupted Oracle. All of the big tech incumbents today also have lots of lots of older product infrastructure (Maps for Google, Dynamics for Microsoft, Mac Pros for Apple, etc) that they must maintain which is also difficult and risky to eliminate. But that infrastructure also imposes a drag on development for new features and capabilities. It is possible to build something simpler and smaller with higher profitability than Google Search (as one relevant example).
Smaller and simpler I buy. Higher profitability I don't. Every serious analysis of this has suggested that ChatGPT is a loss leader, for example.
I'd similarly be very skeptical of claims that Kagi has better margins that Google Search (and I doubt even the Kagi founder would make that claim with a straight face):
Kagi is losing money hand over fist currently, which is why they're moving towards a controversial pay as you go model.
I was the one who made the case to the founder back in 2021, likely along with several others, who didn't think providing unlimited search for a flat rate had any path to profitability.
TikTok is just the current trend. Remember "vines"? I don't think TikTok offers something truly disruptive. They had the right timing and marketing for a specific audience. In a couple of years it will be something else.
TikTok (Douyin 抖音) can just lose money and have it's losses covered by the Chinese Government, as long as it keeps broadcasting the right message abroad.
I personally know someone who used a Magic Eraser to whiten their teeth after watching a TikTok video. That is a terrible idea for so many reasons. Messages like those possibly benefit America's enemies by weakening the populace, whether through sketchy medical advice, promoting risky behavior like challenges, or suggesting illegal activities like hot-wiring cars.
Your suggestion is that the Chinese government is subsidizing tik-tok in order to increase the amount of crackpot medical advice and other scams Americans are subject to, in order to weaken Americans?
I am pretty confident Americans are quite capable of selling plenty of crackpot medical advice and monetizing viral scams to ourselves on our own, thank you very much. It may be something we are in fact #1 at, with no sign of diminished capacity.
TikTok is disruptive and has replaced other disruptive technologies. And I'm sure, that not a few computer nerds are busy programming the next time and life sink.
It's a competition for how much man is willing to flush his existence down the toilet.
ChatGPT feels like a good example, I think 6 months ago that those who weren't in the know would be able to predict something like that. I know I wouldn't be able to, and I even build some small neural nets for fun and read Neural Networks and Deep Learning [1]. I am by no means an expert in the field, but I know some basics and have programmed in all kinds of languages, watched 2 Minute Papers quite a bit and yet I didn't see this coming at all.
That and stable diffusion. While I am not really excited about either technology use in my personal life, I do understand what a huge leap they are. They are both things that few saw coming in the fashion they did and so quickly.
Yes, it took decades of research to get the foundations made and then all just sort of suddenly crystalized into a cogent thing.
While there are folks out there who could point back and call it now, a lot of them were just throwing mud at the wall hoping it would stick. Like Jeremy Rifkins predicting stuff like this back in the 90's to put us all out of work by the far off year of 2000.
What will be the next big thing? I haven't a clue - few people probably do.
This is an interesting re-look because in '23 we are now starting to see what the next big wave could be, and it is obviously generative AI. If I took the diagram on the blog post and extended it, it would be something like the interaction model today being discrete and human-powered, but now we can envision a future where poking at things becomes an outdated model and computing comes to us. Voice assistants were a misnomer, they were voice UIs, but LLMs could potentially be the assistant we never got from Alexa, and transform how we use computers entirely.
AI models might transform how we use computers, but will it also change the outcome? Will it make society more productive, efficient and generate more growth? Or will it just provide more forms of entertainment, maybe provide some convenience. Or worse make the bureaucracies of life even more impenetrable?
Don’t know the answer, but as with software/internet in general we need to at least watch out for false sense of progress.
I personally hope the next big wave will be somewhere along the IoT line and physical automation, agri-tech, health-tech, where software transcends the virtual and finally start automating the physical.
It's really too early to tell what the long term impact of LLMs will be. I know it's new, and exciting, and so on, but we saw the same hype with VR, with 3D TVs and so on.
It's perhaps unfortunate that the high-profile one is made by a company with AI in the name. This is confusing media, and by extension, non-techies into into using the word "intelligent".
Sure these models pass a Turing test, and they write incredibly well, but they're not intelligent. They're basically a search engine with good grammer.
As humans we consider well-spokeness as a proxy for intelligence. How we mock those who speak broken English - they are clearly not intelligent. By contrast the well spoken person is clearly smart (even if what he says is bullshit to those who actually know.)
So here is a computer that speaks with excellent grammer. Clearly it is intelligent. Clearly it is out to rule the world.
The truth alas is less exotic. By scraping a zillion Web pages, and reading a zillion twitter feeds, it merely parrots what has already been written. Computers used to be good at math (once a sign of intelligence) now they are good at language as well.
Is this useful? Sure. There are lots of places where good writing is important. Is it the next wave of computing? It's too early to tell. I have my doubts, but that I'd likely somewhat fueled by the many waves I've seen on the horizon that were ripples when they got to shore.
I've played around with ChatGPT a fair bit. My feeling is that I can imagine it for high school essays or low-effort corporate BS.
It might also be useful for someone looking at a blank sheet of paper who really struggles with that. I can even imagine a degree of generating a template for certain types of articles. OTOH, editing from an an existing article can actually be harder and produce lower-quality output than starting at said blank sheet of paper.
the hype came so fast with LLMs... it has me sus that it's actually the next big thing... then again I hated the ipad... I think I was right on it long term but also wrong on my positive view of humanity (humans aren't just consumers, why use a consume only device). the internet was/is huge, it took forever for it to become hyped... tho the internet is a hype accelerant. Docker is big, and slow hype. Crypto was actually slow now that I think abput it but it went absurd and died recently (queue explaints on why the blockchain is better than functional programming). Crypto wasn't baked before it got hyped I think is my theory, LLMs feel similar. But we'll see. Maybe I'm just old. I've tried chat gpt for a few months and I've already burnt out on it.
From mainframes to everything has a microchip. Maybe the next big era will be AI, maybe biotech, or space :shrug: I'm sure it will be probably something that even people that grew up with computers will find pretty weird, like my grandpa figuring out computers.
i basically disagree with the point that the phone is more important than the personal computer. The phone you have eveywhere, yes, but it is a worse experience, and the main thing keeping them "dominant" is the business interest in preventing you from using adblock.
In a well regulated regime, I cant imagine people using a phone when you are at home and have access to an actual computer.
> the main thing keeping them "dominant" is the business interest in preventing you from using adblock.
You might be right, but I doubt it.
Phones allow people to connect with each other. Smart phones bundle many useful things in a small package (camera, music, communications, games, productivity, ...) in a way that a computer cannot. Mobile matters.
I have every single one of those things in a <3 lb laptop, and the experience of using them is between 2 and 100 times better depending on the category. I'm confused. Is your claim that the way that they are bundled is different from the way they are bundled on a laptop? Because the only way in which this is true is that I can't block ads in music or play whatever game I want on my phone.
I think you'd probably be surprised how many (especially young and older) people are just fine with using a phone for basically everything.
>I cant imagine people using a phone when you are at home and have access to an actual computer.
Even I do this and I have a spare laptop open on the dining room table downstairs basically all the time. But if I'm sitting on the couch and just want to look up something quickly, I'll absolutely use my phone. (Or my tablet but that's the thing I actually don't use much at home.)
ADDED: Which is more important? That seems like fodder for a good faux-debate. They've both been important but for different reasons.
Cars achieved stability because roads and cities did. Functionally the demands on cars have not changed in 80 years; since then it's mainly aesthetics, ergonomics, and economics.
However, for computer automation there are an unbounded number of new kinds of things to do, many unlocked by prior breakthrough's. With everything that interacts or moves running software, there's plenty of room for innovation.
DTC ads were never innovation; they've always been a way to pay the bills, like printer ink. Cross-subsidizing gargantuan organizations has never been a good way to spawn innovation, and it's only worse when the real money is in consumer data or surveillance.
51 comments
[ 3.2 ms ] story [ 105 ms ] threadIt's also more than a little dystopian as the corporate stranglehold over all of us tightens just a bit more.
Today there is the choice of a large variety of safe, efficient and performant car models from insanely low prices (once inflation is taken into account), and soon there will be a similar range of choice in EVs.
Generally speaking, quality is vastly higher than it has been in most lifetimes too - when did you last see a new car with rust on it, for example? (A few brands, notably Mercedes Benz, produce far worse products than they did in the 90s, but they are the exception).
If a wide choice of quality products is the future of the software industry, I'd welcome that, since it certainly is not the state of it today.
For instance, there are aspects of new cars that are ubiquitous and I find so objectionable that they prevent me from buying one, probably ever. If the market were more vibrant, I suspect that I'd have real options aside from "I can't have a newer car".
Also, oligopolies don't automatically mean that the products are improved. Usually, it's just the opposite.
I suppose how true this is depends on what sorts of options you're looking for. The number of acceptable cars on the market now, according to my criteria, is lower than it has ever been.
It comes with industrialisation and commodification and norms and regulation, so it's not all bad either as it consolidates/improves the general quality and availability.
(cue Dire Straits' Telegraph Road)
On the flip side, I can almost see part of your argument. The weather is a perfect example of your argument. The US government funds all the weather models out there, but due to lobbying, they're not allowed to actually have a website that tells us the weather. Instead we get like 5 crappy ad-infested sites that are terrible to use and just make a few people rich. It's insane.
Really? Because weather.gov has showing forecasts since 1995. You can view SF here: https://forecast.weather.gov/MapClick.php?textField1=37.78&t...
How would it be more efficient without a way to know if the end result is any good?
Sure, they got too big for their britches and created propaganda campaigns against pedestrians. They set up a status quo re: the environment that future generations will hate us for. But then their ability to be evil hit a plateau. The cancer that ought to have killed them dwindled, and for better or worse, we collectively said "this is fine". As the head of a car company, your paths to god-king-of-the-universe are significantly limited at the top end, and therefore so too is the extent to which your power turns you into an ineffective lump.
Information companies are not so limited. So the cancer that comes with being powerful is much more likely to be terminal. That is, there's nothing stopping Google from driving their user base away by just showing 100% ads and 0% content. They'll get worse and worse until we're all throwing money at anybody who stands a chance at putting them down.
Just look at Elon Musk, for whom efficacy appears to be secondary to simplify being the biggest threat to the status quo. That dynamic isn't going away: When he ceases being The One another anomaly will emerge.
So sure, it'll get worse before it gets better, but it's not going to be permanently worse. Even if we don't shoot them first, the bad guys will shoot themselves in the foot eventually.
I used to think this way, but I have to admit the last decade or two has made me much more pessimistic about this.
There are just too many examples of oligopolies that have never stopped being terrible. I think the auto industry is one of them, and I suspect the software industry is likely to be another one.
Smaller and simpler I buy. Higher profitability I don't. Every serious analysis of this has suggested that ChatGPT is a loss leader, for example.
I'd similarly be very skeptical of claims that Kagi has better margins that Google Search (and I doubt even the Kagi founder would make that claim with a straight face):
https://news.ycombinator.com/item?id=30822830
I was the one who made the case to the founder back in 2021, likely along with several others, who didn't think providing unlimited search for a flat rate had any path to profitability.
TikTok (Douyin 抖音) can just lose money and have it's losses covered by the Chinese Government, as long as it keeps broadcasting the right message abroad.
I am pretty confident Americans are quite capable of selling plenty of crackpot medical advice and monetizing viral scams to ourselves on our own, thank you very much. It may be something we are in fact #1 at, with no sign of diminished capacity.
It's a competition for how much man is willing to flush his existence down the toilet.
http://neuralnetworksanddeeplearning.com/
Yes, it took decades of research to get the foundations made and then all just sort of suddenly crystalized into a cogent thing.
While there are folks out there who could point back and call it now, a lot of them were just throwing mud at the wall hoping it would stick. Like Jeremy Rifkins predicting stuff like this back in the 90's to put us all out of work by the far off year of 2000.
What will be the next big thing? I haven't a clue - few people probably do.
Don’t know the answer, but as with software/internet in general we need to at least watch out for false sense of progress.
I personally hope the next big wave will be somewhere along the IoT line and physical automation, agri-tech, health-tech, where software transcends the virtual and finally start automating the physical.
It's perhaps unfortunate that the high-profile one is made by a company with AI in the name. This is confusing media, and by extension, non-techies into into using the word "intelligent".
Sure these models pass a Turing test, and they write incredibly well, but they're not intelligent. They're basically a search engine with good grammer.
As humans we consider well-spokeness as a proxy for intelligence. How we mock those who speak broken English - they are clearly not intelligent. By contrast the well spoken person is clearly smart (even if what he says is bullshit to those who actually know.)
So here is a computer that speaks with excellent grammer. Clearly it is intelligent. Clearly it is out to rule the world.
The truth alas is less exotic. By scraping a zillion Web pages, and reading a zillion twitter feeds, it merely parrots what has already been written. Computers used to be good at math (once a sign of intelligence) now they are good at language as well.
Is this useful? Sure. There are lots of places where good writing is important. Is it the next wave of computing? It's too early to tell. I have my doubts, but that I'd likely somewhat fueled by the many waves I've seen on the horizon that were ripples when they got to shore.
I've played around with ChatGPT a fair bit. My feeling is that I can imagine it for high school essays or low-effort corporate BS.
It might also be useful for someone looking at a blank sheet of paper who really struggles with that. I can even imagine a degree of generating a template for certain types of articles. OTOH, editing from an an existing article can actually be harder and produce lower-quality output than starting at said blank sheet of paper.
In a well regulated regime, I cant imagine people using a phone when you are at home and have access to an actual computer.
You might be right, but I doubt it.
Phones allow people to connect with each other. Smart phones bundle many useful things in a small package (camera, music, communications, games, productivity, ...) in a way that a computer cannot. Mobile matters.
I have every single one of those things in a <3 lb laptop, and the experience of using them is between 2 and 100 times better depending on the category. I'm confused. Is your claim that the way that they are bundled is different from the way they are bundled on a laptop? Because the only way in which this is true is that I can't block ads in music or play whatever game I want on my phone.
when you are going on a run, or a bicycle ride, or you're lying on the beach or take a bus, you take your laptop out to listen to music?
...
Mobile would not have ruined our Internet and GUIs and psyche, and people would probably not take pictures anyway and live in the moment as a bonus.
>I cant imagine people using a phone when you are at home and have access to an actual computer.
Even I do this and I have a spare laptop open on the dining room table downstairs basically all the time. But if I'm sitting on the couch and just want to look up something quickly, I'll absolutely use my phone. (Or my tablet but that's the thing I actually don't use much at home.)
ADDED: Which is more important? That seems like fodder for a good faux-debate. They've both been important but for different reasons.
Cars achieved stability because roads and cities did. Functionally the demands on cars have not changed in 80 years; since then it's mainly aesthetics, ergonomics, and economics.
However, for computer automation there are an unbounded number of new kinds of things to do, many unlocked by prior breakthrough's. With everything that interacts or moves running software, there's plenty of room for innovation.
DTC ads were never innovation; they've always been a way to pay the bills, like printer ink. Cross-subsidizing gargantuan organizations has never been a good way to spawn innovation, and it's only worse when the real money is in consumer data or surveillance.