I think it really was the M1 Chip that killed Mighty.
The product itself looked like it was working great.
Before the value proposition was good.
You can still use your old laptop and still have a very fast browser experience via remotely running it on a AMD server chip, this way you save on battery life and RAM too.
Then M1 speedometer scores started beating AMD Epyc on single core web browsing performance. Not only that, but even providing a much better experience in terms of battery life because of power savings.
Then yeah there was no point paying 20$-30$ a month for a remote browser instance. If you can afford that as a user you're either already set on CPU/GPU/RAM or better off just buying a used m1 air for <900$.
It's quite sad of a lesson where you do everything right, work consistently hard, be quite innovative, have plenty of financial and talent resources, and still fail.
> I think it really was the M1 Chip that killed Mighty.
Mighty is only dead because Suhail wanted to pursue what they clearly think is a bigger opportunity [0]. Mighty is already building its own GPU servers, might as well put them to use for AI as the perf and the size of AI models is roughly on an exponential trajectory [1]. It may not be long before AGI is within vicinity [2].
Though I agree a better strategy wouldn't be to accelerate the browser (akin to boiling an ocean?), but instead take one app at a time and make it browser-native (like, Photoshop -> Figma). A similar concept to (I don't remember who said it), take tools hackers use and build it for the Internet (grep -> Google, sendmail -> Hotmail, emacs -> Replit, sftp -> Dropbox etc).
[0] AIs today can automate away humans on quite a few tasks, assist them on quite a few complex ones already. And still: models are getting only way more capable, not less.
everything except for, as Suhail points out, picking the wrong side of a technical trend to bet on. All the technical wizardry in the world can't save you from bad strategy.
> I think it really was the M1 Chip that killed Mighty. The product itself looked like it was working great.
Agree with this take. It's amazing he was able to get this funded at a time it was fairly obvious from their product roadmap and language that Apple was going to release ARM Macs. Props to him for the skill it took to get funding in the door.
>It's amazing he was able to get this funded at a time it was fairly obvious from their product roadmap and language that Apple was going to release ARM Macs
It's easy to forget, but a lot of people were rubbishing the idea of ARM Macs as a viable replacement for Intel until it actually happened.
I did see that opinion expressed, but it was just silly given Ax performance in iPad/iPhone pre-M1. In no world does the flagship laptop use a more expensive, less efficient, and slower chip than the mid-range iPhone for more than a couple of product generations.
I don't think it was that silly to expect a transition from a stable, mature architecture, sold (albeit too expensive) by an established chip maker, to a new (to desktop) architecture designed in-house by a company known for its phones, especially when you consider that desktop workloads are very different from "here is an API for our walled garden" workloads.
Add to that the fact that a successful transition was entirely dependant on how well rosetta works (it turns out it works well, but that wasn't a given at all), evidenced by how people got burned in the powerpc-x86 transition
> when you consider that desktop workloads are very different from "here is an API for our walled garden" workloads.
This is a solid point, but iPad workloads were pretty clearly converging with laptop workflows. (And Apple was obviously moving in the direction of converging the SDKs even before the M1 was announced.) Plus, they already had the ability for devs to build apps to run on MacOS & iPad/iPhone. This was a strategic move that was visibly underway for years.
> a successful transition was entirely dependant on how well rosetta works (it turns out it works well, but that wasn't a given at all), evidenced by how people got burned in the powerpc-x86 transition
All the more reason to expect them to make a transition -- they had pulled it off successfully (if with some hiccups) before. I would not be surprised if they had some of the original Rosetta team around to revise for this time around.
In seriousness, going by what Suhail's been documenting through their tweets, they seem to fit the relentlessly resourceful bracket to a tee, which (according to pg, at least) means they have an above average chance at succeeding doing venture-scale startups: http://www.paulgraham.com/relres.html
I remember people being very impressed with the performance of the arm macs at the time. Maybe they were impressed that it met expectations though and correctly predicting it would be good.
I loved how fast Mighty was (even on M1 it was noticeable) but there's no way I could justify that expense and I'm a big early consumer of B2B SaaS tools.
For Mighty, this would mean their addressable market shrinks every year as more people replace with computers that are faster. That's not the characteristic of a market most startups want to be in.
Even for gaming and workstation use, this seems to be true. Last year, the value proposition of remote desktops looked great, but it turns out price/performance of all computing parts was at an all-time high.
I looked into this and found, annoyingly, that even if used the Remote Desktop rarely and I could deallocate the machine to not get charged, it was actually the price for storage that made it too expensive.
You could say that same about running off-line batch jobs on EC2 instances. Unless your co-located data sources are forcing you to do so, I cannot fathom why you would.
did this product have any appeal? the price was insane for something i get for free, if you could afford it and needed it you could afford a decent computer that could
It did when it came out. But it got already leapfrogged. Calculating fibonacci(45) with a recursive javascript function takes 13s on my m1pro and 7s on my ryzen 7600x. Profiling our angular app in chrome the initial render takes 200ms on m1pro and 130ms on my ryzen 7600x.
Apple silicon is still the master of performance per watt. But raw performance: It really was just tsmc 5nm what gave apple the edge. Apple will shine again when they sit exclusivly on tsmc 3nm, but until then, not so much.
I’ve talked to a couple of very loyal Mighty users. If you think about the number of hours a knowledge worker spends in the browser every month, it works out to a pretty small price per hour.
It’s similar to people in jobs that require them to live out of their inbox paying (a comparable amount) for Superhuman.
I'm really REALLY not sure how it makes sense to pay that much for a remote browser unless there's something I'm missing which I have to think is the case. What was the actual hook here? Surely a remote browser cannot have been worth it for 240-360 dollars a year? Wouldn't it have been more worth it to simply buy an M1 Mac or a Surface? How are people suffering from slow browsers that much?
There's a reasonably sized market segment of people who, themselves, work at startups with dubious value propositions. They try to prop up the whole ecosystem by spending a lot of money on each other. (Source: worked at a late-stage startup, spent weeks integrating APIs that did nothing but were owned by a friend of the CEO.)
There's another segment of people who can be sold snake oil. They aren't actually very sensitive to UI responsiveness, so they won't notice the extra latency added by the streaming service, but they will believe claims that it is "faster" than what they're currently using in some nebulous way. (These are the same kind of people who rewrite their app in React because they believe React makes apps faster.)
> Wouldn't it have been more worth it to simply buy an M1 Mac or a Surface?
Yes, probably, but I would note that a cheap M1 Mac is 4x the annual cost you cited for the browser, and if the browser is 95% of what you use a laptop for, getting the same perf boost for 25% the cost sounds like a good value prop.
Granted, a laptop should last more than a year, but presumably it starts to show it's age in under four, whereas the remote browser can improve over time.
I had Macbooks that lasted 5 years (2010-2015) pretty well, my current MBA M1 is almost 2 years old and I don't see myself needing to swap it for another 2-3 at least. My girlfriend still uses her MBA from 2014.
I don't really see how I could justify paying 20-25% of the price of my laptop per year, that I use for much more than just browsing, to have a faster browsing experience. And I'm a power user (SWE, etc.) compared to the mainstream consumer, the market for this product seems to be extremely niche/small.
Ironically BAAS is sold for security reasons too. As in isolation. You'd do very sensitive stuff with the remote browser, or vice versa, for isolation from other activity.
There was no point paying $30 a month _period_. You could get a cheap, refurbished 2014 laptop and it would still be a better value proposition than Mighty. The only potential buyers of this service are SV tech bros, not anyone living in the real life.
Honestly one of the most "But why?" startup ideas I'd ever heard of. I figured pg's hype for it was based in thinking it could someday be targeted for profitable acquisition by Google. But perhaps the current tech industry downturn (plus Stadia's death) made that obviously untenable.
At $30/month, this sounded like something that could only be worthwhile for businesses (though I'm at a loss to imagine what size/industry of company would want this). Did Might have any corporate clients, and if so, how did they navigate the security/privacy compliance issues?
I think pg likes slightly naughty ideas. Didn't he famously want to program a server using instructions send over email? And this Mighty idea was quite naughty. Stream like Netflix but it is your whole browser. And also pg really hypes YC companies which, bless his heart, is what we all expect and why YC is such a beast.
I obviously don't have numbers at hand, but I'd bet most people have a internet connection with a latency of at least 100ms to closest game service servers, unless you already work in IT or adjacent sectors where latency matters more. Not even 50% of people online in the US are connected via fiber optic connections, so extrapolate that over the rest of the world, although many countries have better connections than available in the US.
There are a lot of sources of total latency in most games and systems that affect how long it takes between when you click a button and get a response. This means that there's often opportunities for optimizations that can remove entire frames of latency to make up for the additional network lag. It's not going to be enough for e-sports games but it can keep the experience from changing for many titles.
Just as an example of what I'm talking about, if you turn on the nVidia Reflex latency reduction feature along with the DLSS 3 AI Frame Generation, you get the same latency as with Reflex off but with much better smoothness. Of course if you're playing Overwatch you are chasing every millisecond of latency, but for most games it just needs to be low enough not to be noticeable.
Only 2% of the US population owns a Xbox Series S/X, 2.3% own a PS5.
The average internet speed in the US in 2022 is 120mbps.
Asking people to have a >250mbps connection to using cloud gaming vs being one of the most popular gaming consoles in the country isn't much to ask... in order to be a moderately successful game platform.
>I've played a 50hrs of Xbox cloud games in the last 2yrs and never had latency issues
Gonna be honest , I'm not taking gaming advice off someone who has played only 2 hours per month.
Especially with regards to something like input lag.
I don't play competitive online high framerate FPS games where that level of input matters. If I did I would invest in a windows PC.
But for all of the single player games I played it was never a noticeable problem. If you're going to complain about anything with cloud gaming it should be switching or starting a new game. The loading screen preamble before the game start screen is the only annoying part. But that doesn't affect gameplay.
If input lag was such a real problem I highly doubt Xbox would have redesigned their new 2023 Xbox UI to be 100% centered around Gamepass (The current one already pushes it hard but the new one leaked on Youtube makes it the near default).
Props to Suhail for building in the open (takes a lot of rejection tolerance) and being so honest about the path.
That said, I work with lawyers as customers and I suspect there was an adjacent market in data privacy that a hosted browser would have had PM fit for.
A lot of industries have strict legal requirements on controlling employee access to data (think healthcare, legal, compliance). In these cases, SaaS becomes risky if it can be accessed by anyone off their work computer.
The standard solution is to limit the webapp to be accessible on VPN, and limit VPN to be accessible on MDM controlled devices, and limit MDM to be accessible on company owned devices.
It would be a lot easier to just control the browser viewport and prevent data harvesting (essentially what Mighty was doing, focused on performance).
Sounds like a small change, but it makes a world of difference when it's not feasible to send a physical machine to someone (3rd party contractors, overseas employees, low wage employees, etc)... something I discovered when talking to lawyers about using contractors overseas for routine data entry tasks.
But that's definitely not the same kind of user, or tech expertise and would have been a big pivot in itself. Hopefully someone buys the IP and builds something cool.
As evidence of a market for this, see the company Citrix. Also I think Facebook's content moderation website is only accessible from some sort of remote desktop system.
Funnily enough, the few times I've seen Citrix remote-desktop browsers used it's been for the opposite use case: isolating potentially malicious websites in a virtual machine so they can't compromise employees' computers.
> The standard solution is to limit the webapp to be accessible on VPN, and limit VPN to be accessible on MDM controlled devices, and limit MDM to be accessible on company owned devices.
Didn't the "BeyondCorp" zero trust model pretty much kill that, or at least show there was a better way to restricting access to secure apps than a VPN?
BeyondCorp canonically includes device authentication! The way I’ve seen it implemented is a browser client certificate though, kept valid by MDM. No need for VPN.
Of course a certificate could be stolen/transplanted but you would need to compromise the laptop first, and that’s also true of VPN solutions unless the keys are in TPMs.
And yet security VCs are investing here again and yeah, no idea why This Time Is Different
We do get VDI users for our tool in some high-end security sensitive places to work around weak clients (budget is not uniformly distributed across users), but that niche is a small market wrt VC..
> It would be a lot easier to just control the browser viewport and prevent data harvesting (essentially what Mighty was doing, focused on performance).
Not really, the client can still OCR whatever the browser viewport contains, even if the server just streams video.
I love the vitamin vs painkiller discussion. The basic premise is: don’t build a vitamin, build a painkiller because it’s solves a real problem and people will be willing to pay for it.
First of all the analogy fails in the real world. The vitamin market is huge. People pay lots of money to buy vitamins.
Second, the transition is not clear. It’s hard to draw a line line and say now this vitamin turned into a painkiller.
Third, I believe almost no product is a real painkiller. We in the West at least live in sheer abundance. Almost none of our problems is really painful. For example: was Facebook a painkiller? Hardly. Was the iPhone a painkiller? Not until you considered using it.
I’m not saying the analogy is useless. But I don’t think that vitamins can’t be super successful.
Analogies always fall apart somewhere under scrutiny. This one is still useful though to compare. Vitamins are successful because there’s a lot of awareness and marketing (sometimes deceitful) about the benefits of vitamins. If you’re going to build a vitamin product be prepared to spend a lot in marketing and/or have a killer user experience with tangible results to be able to rely on word of mouth.
> Analogies always fall apart somewhere under scrutiny.
Yeah but this one falls apart the moment you simply try to figure out what facet is trying to be conveyed. Just say what you fucking mean. The general painkiller > vitamin argument is generally bunk/false dichotomy - therefore you’re not saving any time with the analogy.
It’s not that it doesn’t click - it’s just another stupid thing like Myers-Briggs personality types - and it’s just a shit analogy. Go ahead and categorize businesses by vitamins, pain killers and candy if it’s makes you happy. I believe reading anything into these distinctions is open to the free interpretation and is just astrology for the tech bro crowd. And no one actually agrees on the categorizations for anything less trivial than Candy Crush - as evidenced down thread. The software businesses that are cleanly categorized by this are the exception not the rule.
Once you start having a meta discussion about the analogy itself, perhaps that should be a sign.
Imagine the misery people went through for centuries prior to the discovery of pain killers. If you’ve ever had a tooth ache, I’m sure you can attest to the efficacy of pain killers as a medical invention. They are an enormously important piece in the quality of life that most of us enjoy today.
The vitamin vs pain killer metaphor is about building a company with a defensible business model.
So, the metaphor is about arriving at an honest answer to this question: would the “quality of life” of your first customer improve significantly if they switched to your product?
If you answer is no, you have two options:
1. either make changes to your product to make it more attractive to your target customers or;
2. keep the product as-is but change your positioning so that the product can be marketed to a different set of target customers to which the answer to the question is yes.
> First of all the analogy fails in the real world. The vitamin market is huge. People pay lots of money to buy vitamins.
Also in the real world, people who sell vitamins lie a lot, and some people believe some really stupid things about vitamins. So the ultimate vitamins are antivirus products, I suppose: Sometimes needed, often a waste of money, and actively harmful more often than the naïve would suspect.
I like the analogy. I never heard it before now, but I instantly understood what I think it meant: build something that is harder to live without. I have nothing to back this up but I’d think most people think of a vitamin as easier to live without than a painkiller.
> For example: was Facebook a painkiller? Hardly.
I don’t know. People without pain do take painkillers. And they can have a really hard time stopping. In that sense, I would argue that Facebook is a lot like a painkiller for a lot of people!
> I don’t know. People without pain do take painkillers. And they can have a really hard time stopping. In that sense, I would argue that Facebook is a lot like a painkiller for a lot of people!
Nice, you’ve explained by demonstration better than I could why this is a trash analogy and only confuses issues, does not clarify them. Consider stopping using it.
The reason it’s breaking down is you’re using it out of context — at least in my understanding. It is primarily a B2B analogy, and you’re talking about consumer tech.
In B2B people buy painkillers because then they don’t have to do that painful part of their job. They rarely buy vitamins (although some do) because those might require them to do more.
It’s fundamentally about outsourcing/automating your own functions.
Vitamins can def be super successful, as you mentioned. Agreed.
I think that with Mighty it's just the combination of being an expensive product (mainly to operate, but to some extent for the end customer as well) + privacy concerns + most people don't suffer that much from slow browsers and instead willing to pay $20/30 a month to solve it.
Pretty niche market, I think. But might be mistaken...
My pain point was that my friends were fragmented among MSN, AIM, Skype, Google etc. and I had to pay for texts so I always used online messaging instead
When I was able to add everyone on Facebook I was able to keep track of everyone, and also connect to all the XMPP networks.
Currently it doesn't serve this purpose because it's no longer a universal messenger, so I don't go on it
Well I think M1 probably put the nail in the coffin but I agree, probably still fails anyways. Mighty is cool and useful, but is it cool and useful enough for the non-trivial server costs? $20-30/month is a lot for a service.
I am very skeptical by my nature, and see it more as a negative trait. But there are moments when I want to shout from the top of my lungs: "Who in their right mind gonna use it, not talking about buying it?!". I remember two such cases: that voice rooms app, where people gathered and talked about themselves and of course crypto tokens, and this remotely running browser for $30/month.
I beleive, by the way, that the idea of a fast browser for cheap devices came to founder after reading story about Browserling, QA multibrowser suite, going viral in India, where people with $20 phones were able to use WhatsApp webapp via Browserling free plan (HN post: https://news.ycombinator.com/item?id=16103235). Coincidentially, I heard that voice room app (I am not being sarcastic here, I really forgot it's name and don't think it's worth googling) also became succesfull in India.
Mighty had no hope of succeeding. I’ve built something very similar to Mighty, and the fact is that browsers are hogs, and then adding on video transcoding on top of that makes the situation completely untenable. If someone can’t afford a computer good enough to browse the web, they sure as hell can’t afford a computer that can transcode HD video at all times, which is essentially what they’d need to pay for.
This was EXACTLY the problem Mighty was encountering.
They were targeting customers who didn't have powerful enough computers and offering a way to make their computer effectively faster while browsing the web without needing to upgrade their computer.
The problem is the solution costs $35/mo. That's $420 a year.
Anyone who can afford an extra $420 a year to improve their computer speed, probably is just going to buy a better computer with that money.
Browsers aren't single threaded, so the ST performance isn't the end-all-be-all metric to use. JavaScript is single threaded, but the browser itself uses multiple threads. E.g. there is the main thread, which handles the user input. There is the rendering thread, which actually takes up a majority of the CPU time on a lot of websites. And then there is stuff like GPU acceleration, which makes a lot of the web performant when it wouldn't be otherwise. Of course, a server does all of these things worse than a decent laptop.
It is the end all, be all metric. The reason is because once a certain threshold of MT performance is reached, which the M1 most certainly reaches, then the ST is what matters for how fast your website loads, even with multiple tabs opened.
Predicting failure is easy, though, because most things fail: you can predict everything will fail and be right 90%+ of the time. The venture capital model operates on that understanding: most things will fail, but in taking a chance on lots of things that will probably fail, you're exposed to the very small number of things that will succeed in a meaningful way.
Success is the aberration, and being able to identify the things that will succeed is where the challenge is. If you think you have talent in identifying things that will succeed, then invest in things, and if you're right, you'll make lots of money. Unfortunately, as you'll discover, predicting failure is easy because there's a million ways to fail and very few ways to succeed.
Mighty investors would have had 90%+ confidence it would fail, but that's fine, because it's the tiny chance it might succeed that mattered.
You're talking from a pure statistical standpoint. Yes, the probability that a random startup fails is higher than the probability that it succeeds. Nobody denies that. I doubt anybody said, "I predict Mighty is going to fail because most startups fail."
It's much better to argue on WHY people said that Mighty was a dubious product proposition -- from a first principles perspective. That way at least we can learn some lessons from it. Contrary to what people are saying, I doubt the reason for failure is that since Mighty's inception the browsing experience has gotten much faster. Because it really hasn't.
Mighty was as doomed in the beginning as it is now. -- and you know it if you've been using a web browser for the past few years. If you don't live in a bubble, you realize $30/month is absurd. These are the important points, not "most startups fail."
Nah, predicting failure or success is exactly the same thing.
VCs don't simply predict success and bet; they use their power to make it happen or crush it.
Overall, I have a very low confidence level in most of the stories told by VCs about themselves, and I am not sure about their positive influence in the economy or technology.
You can’t short bad ideas directly. But you can (or could until recently) make money off them by coming up with them and pitching them to VCs yourself. That’s kind of like shorting.
That's a terrible bet to take as a founder. Assuming even odds, put aside $10k for some time, and you may double it if you're doing well and probably don't need it, but if you're failing, you lose the money.
> I wish there was a way to make money by shorting bad ideas. I think I could easily fund a few companies by doing that on weekends.
No shit? The vast majority of VC funded ideas fail, so you'd make money on almost every trade.
The issue is that your fund would blow up every time you shorted the next Airbnb/Uber (two companies that everyone agreed were terrible ideas at inception)
This is why general prediction markets need to continue to flourish. They're an antidote to bullshit. Also, ban the bettors that things wrong because they're probably just gambling addicts.
Looking at tweets like [1] it seems like their original bet was that server single core performance would dramatically outpace client single core performance. Honestly even if that had happened I still don't think their product would have made sense. The hosting cost and bandwidth/latency requirements would still have killed it.
The death of this product is a good sign for computing overall. The more of computing that happens on the client the better. Clients are (at least sometimes) under user control. When stuff moves to the cloud users give up control and that's a bad thing. Long live fat clients!
The product, "compute pixels on powerful machine -> render on thin client", makes much more sense if you don't try to monetize by being a cloud landlord, but you let the users own that powerful machine, make money through hardware, and help them with the software, make money through support, no subscriptions, no rent-seeking.
Think at the level of a family: 2 parents, 2 children which means: 4 smartphones, 2÷3 laptops, 1÷2 desktops, 1÷2 tablets, each of them $1,000+ which means around $8,000+ and $11,000+, replaced every 2÷4 years. Instead of 8÷11 screens, each with it's own beefy CPU/GPU/RAM, you could have 8÷11 thin clients (at the most $200 each, for fancy cameras and larger batteries) and a powerful machine, somewhere between $5,000 and $10,000, dual CPU, dual GPU, etc., which can last 10+ years (thinking of all those refurbished servers with 2012 Xeon E5-2620s which work great even today, and have no reason not to work perfectly fine even in 2032).
Bandwidth/latency is an issue but as we are advancing in 5G+ technologies it's just a wait game.
Long live thin clients where the user owns the upgradeable fat server rendering the thin client!
I think something like this could make sense as a place to run large machine learning models locally. I don't need a server to run my browser or my apps, they run just fine locally. But a server to run large language models, Stable Diffusion, Whisper voice recognition, etc would be useful, as these types of models can run much faster at higher quality on a beefy GPU than they ever will on a phone.
The endgame of these models is an agent that knows practically everything about me and can perform tasks on my behalf, which I would really prefer to live in my house running on hardware I own, rather than in a data center under someone else's control.
Precisely, data will become the ultimate material, from chemputers [1] to neural laces, it's all data, and having the user be the true absolute owner of it, as in actually owning the database, is paramount for any sane future use and development of technology. More thoughts and some code on this [2].
After the user owns a server capable of storing the databases and the function execution contexts (see deserve functions) for all their services, having also pixels rendered there and sent to thin clients it's just a matter of convenience and security: no more worries that the device you carry all day can be broken or stolen, it's just a low-cost piece of glass with an uplink to your server.
I find it deeply unsettling that you'd consider it normal/common to upgrade 4 digit (is that all Apple?) devices every 2 years, or that folks would need so many different form factors in the first place.
Just thinking of the literal mountains of ewaste gives me goosebumps.
It's pretty much what happens in the regular household with median and around income [1] [2].
Yes, we are currently generating way beyond ridiculous levels of waste, electronic and beyond: we throw away globally 1.4 billion tons of food per year [3] in a world with 822 million people suffering from undernourishment [4]. If we are this bad with something as vital as food, of course we are worse with devices.
The lower limit of the example budget above was $8,000 and considering a family of 4 over 2 to 4 years.
"Apple generated $365 billion revenue in 2021, 52% came from iPhone sales" [1] Taking Apple as example, those billions have to come from someone. The greater point was that instead of having the billions try to make every single device the most powerful CPU/GPU of the current Moore's Law peak, because no one wants a slow device, you might just as well have a bare glass able to render pixels computed on a remote powerful machine, but that machine must be owned by the user. And you don't stream an app, a browser, you stream an entire OS, on the spot, on request.
Picking screen metrics from 4-5 years ago is a pretty terrible, metric - I have maybe 14 screens but only 2 or 3 of them see active use or are hooked up to anything.
My oldest (working) devices are already all closing in on or past a decade. This metric is flawed in many ways...
"Adults spend $1,200 a year on consumer electronics and own, on average, 25 CE products per household, according to a study by the CEA." [1] In a family of 4 [2], this can mean somewhere around $3÷4,000 per year, or $6÷8,000 every 2 years.
I can think of quite a few electronics I spent money this year which did not include any screen device. Again, these stats are misleading, maybe dishonest.
My "device" budget we'll call it has been less than 1000$ over the past 5 years, and that is with spending a fair amount on non-screen decives, which are electronic in nature but not investments in dedicated computing like you would have us beleive.
"Consumer electronics" consist of "telephones, televisions, and calculators, then audio and video recorders and players, game consoles, mobile phones, personal computers and MP3 players", "GPS, automotive electronics (car stereos), video game consoles, electronic musical instruments (e.g., synthesizer keyboards), karaoke machines, digital cameras, and video players", "smart light fixtures and appliances, digital cameras, camcorders, cell phones, and smartphones", "virtual reality head-mounted display goggles, smart home devices that connect home devices to the Internet, streaming devices, and wearable technology." [1] Most of them have screens.
As a general heuristic, one single person's decision tree is rather irrelevant in the long open game of average spending behaviours.
"one single person's decision tree is rather irrelevant in the long open game of average spending behaviours."
Sure but collectively is what matters, you can't make such precise arguments from such general behaviors, as you put it it's not about a single person's decision tree.
Your own metrics disagree with your original point.
My old gaming computer from 2013 still plays 2022 games on medium-high settings just fine. You can make computers run for a long time if you put in the effort.
Precisely, the users, by definition, care only about the user interface. The matmul machine can be hidden in any box, just that so far it has been tied in same box with the screen. Hence why the greater point is that thin clients are relevant if the user owns the matmul box.
Most people who upgrade on a cycle also sell on a cycle - if you buy the new hotness today, and sell it in two years, it'll still be relatively hot. The old one isn't thrown away, it's just sold to someone else.
People do it with cars (even if they're not leasing, which is explicitly doing it).
And even among households that do have so many devices, aren't the kids' devices typically "replaced" with hand-me-downs from the grownups rather than with totally brand-new devices?
It was just a bit of napkin mathematics seasoned with some quickly searched statistics of averages. The greater point is that thin clients become truly relevant when the user also owns the rendering machine, think of self-cloud.
"It also has other uses in a variety of specialist contexts." [1] Language is complex and signs often have more than one signification, but one must have curiosity and endeavour beyond what stands prima facie.
With this specific meaning of "range between"/"closed interval" it is often used in Eastern European italo-/franco-phone languages/texts. A short history of obelus on StackExchange [1].
yes! this is the same reason I’m glad Stadia died. “cool demo” does not equal “practical business”, and local compute will always be superior to cloud compute.
What? Stadia worked great (I say this as a user). What didn’t work was the business model, the lack of games, and the lack of trust in google to invest to fix those other things.
It worked flawlessly for me for years over Wi-Fi. GeForce now which I currently use is rockier due to being more bandwidth sensitive and not as controller friendly since you’re really using a PC, not a console as stadia was.
Even under ideal conditions where most people were claiming it's working great, I never found the latency even remotely acceptable. Maybe it's due to my preference to esports titles that requires all the responsiveness possible, but the point stands. Remote rendering would never be able to achieve anything even close to acceptable performance to be competitive with on-client rendering.
Agree with co-commenter, I have never seen anyone who used it for low latency apps say it was sufficient. The target market for displacing e.g. gpus is heavily geared towards low latency.
It might have been OK for mobile games, but then they were just competing with themselves, with an arguably inferior product (if I can't play my mobile game anywhere, only in the vicinity of my house where I already have my high powered rig, what's the point?)
I do think it was an attempt to defeat a very legit, widespread use case to try tk justify pushing thin clients on everyone, glad it failed.
My primary feeling is that I think technical people drastically overestimate the pain that most "normal" people experience when it comes to browser performance.
I see it all the time on HN, e.g. people bitching about the amount of RAM Electron apps take and the like. Who cares? The average user certainly doesn't.
If anything, 98% of the time when I experience browser performance issues it's network-latency related, or the fact that some page is loading 300 ad-tracking scripts and one of them is accidentally blocking. The only time I really notice client-side execution performance is when someone posts a cool 3D browser example on HN and things slightly slow down a bit on my phone when there are a couple million polygons or whatever. Even then, things are fine on my laptop.
100% - people cared about browser performance and usability until Firefox and Chrome solved 99% of the problems users would ever face (tabbing, consistent rendering, security, beauty(!)) and now it’s a market that’s immune to change.
My only use for chromium is youtube. I only watch a handful of channels but things are getting ridiculous. Sometimes I get ads every minute. Every minute! Just the other day I got served 5 ads in a row after a simple fast forward. I gave up TV more than 20 years ago and this crap is way worse.
What can I do without letting them in my actual browser? These people are crazy.
This, as much as people may moan, is the answer. Either that or go subscribe to Nebula or whatever not-YouTube platform your favourite YouTube channels are publishing to.
Content production costs money, in some cases vast amounts of it, and while advertising is a terrible solution to that just not paying is an even worse one. If everyone blocks ads then nobody gets paid, and nothing gets produced.
Its not just about moaning, it's about paying it forward. I'm not sure when people decided everything should be free - under a certain income sure we should be subsidized, paying for things is a way of voting for the way you want to see the world.
If you don't pay you don't have a choice. "Not using" is not enough, because you still need to support what you do use.
I pay for premium but still have to use vanced to avoid the sponsor segments that are in literally every video now. I totally understand why people wouldn't bother with premium until youtube offers a native sponsorblock-like alternative
>I see it all the time on HN, e.g. people bitching about the amount of RAM Electron apps take and the like. Who cares? The average user certainly doesn't.
It is our job to care so non IT people don't have to.
I think you got your greater than sign the wrong way around? Consider a lot of non-tech people don't constantly upgrade, and still rock 4GB systems, at least an eighth of that resource is a big ask for a sloppy app (unless it is the only app they use on the device).
I am an insider to this space and I think you are probably an outsider.
That link doesn’t say what you think it does.
LCP before optimization is a cause of what? For example, a complex product page. Well of course that will convert worse, because the product is complicated.
But not too simple! 0.5s performs as bad as 2s.
Anyway, in my real experience, conversion fluctuates so much across products, the range greatly exceeds the impact measured by aggregate conversions by LCP, so your interpretation is wrong.
98% of performance problems on my machine are traceable to a crappy Electron video communication app (skype) eating 4 cores and still having hiccups galore. Switching to a different app usually fixes the issue.
I'm pretty sure the average user cares about "hey this app sucks, but this other app gives me the same functionality without issue".
You can hope they never discover other apps for work-from-home, but frankly, if that's your business model, you're doing it wrong.
> My primary feeling is that I think technical people drastically overestimate the pain that most "normal" people experience when it comes to browser performance.
Where have you lived? Asking because “normal” varies a lot by geography.
> If anything, 98% of the time when I experience browser performance issues it's network-latency related
I guess if you are on HN, you wouldn’t be considered “normal”?
"I think technical people drastically overestimate the pain that most "normal" people experience when it comes to browser performance."
Eh I think technical sales people overestimate how much users don't care, the feeling of "normal" users don't care is primarily driven by a narrow age bracket of folks who haven't had enough time to form an opinion (ages 20-25).
As this age bracket thinks of itself as "normal" and tends to live in its own bubble, it dismisses the general distaste for technology as being "not technological", when in fact the older generations already have expectations which it recognizes that technology is getting worse, not better.
I think this is true in one sense and false in another.
It's true because users often don't complain about or even consciously notice poor performance. So if you are trying to sell a product, advertising performance benefits doesn't work well. People don't think they care (except in extreme cases).
But it's false because even moderate performance differences actually have enormous and easily measurable effects on user behavior. So if you care about providing value, then performance is one of the most important things, and it's consistently overlooked because it doesn't sell.
People complained about the amount of RAM NeWS apps took up, too. In retrospect we realized NeWS was a missed opportunity to revolutionize networked-app UI.
Electron is the dream of NeWS for current-era hardware and software. In the future we won't even blink at the space it takes up.
It was one of the competitors to X back in the day. Basically all the window painting and client-side logic would be handled by an extended PostScript interpreter, which would receive and run the client code from the application, much like we do with HTML/CSS/JS today.
I was skeptical about Mighty but seeing gaming platforms moving to streaming, PG endorsing this as the future, I thought I was just not getting it.
Maybe I should trust myself more.
Stadia failed too… but was that just google pulling the plug, or the whole concept that doesn’t make sense?
Personally I’ve tried stadia and ps now, and they both have noticeable artifacting, lower resolution, and input latency, on a 1 gig symmetrical fiber connection. The only streaming that’s remotely acceptable as a gamer is steam in-home streaming over a hardwired connection.
I think Stadia failed because of a combo of reasons. Main 2 being: 1) Google. 2) Their pricing model made no sense.
Geforce NOW (NVIDIA's Stadia) works a lot better IMO, and doesn't require you go through them to acquire games. It lets you stream games you already have access to on say Steam or where ever.
I agree with the idea, but it may depend on how the X11 application was written.
My experience with X11 over a network (across a city, geographically and distance-wise) was very poor. Slow updates, just extreme latency in doing anything. Not impossible to use, but definitely unusable by any reasonable metric. We're talking dozens of seconds for most GUI interactions.
I eventually piped it through RDP (mRDP, iirc), which itself was piped through a VPN. This vastly improved performance and made it so that I didn't have to be physically present at my workstation. Just as well, since it was during the early covid timeframe.
Ye. I don't think X11 servers were designed initially for anything but very local network connections. Maybe your packages were reordered or something which X11 maybe could not handle very well. Otherwise I don't understand how piping would help.
Yeah it's almost like people are collectively waking up to the fact that VCs are not actually that smart because their whole business model is predicated on being wrong 90+% of the time (which to be fair they've been explicitly saying forever). Now they're connecting the dots and realizing that the other 10% are just luck (and a well timed bull market).
broken clock is right twice a day. being skeptical on 100% of startups makes you right 90% of the time but doesnt by itself prove you have good judgment yet. you also need to know how to be nonconsensus and optimistic.
I never understood the value proposition here, but I assume they had _some_ traction which motivated them to work on this for almost four years. Curious as to their numbers and what people walked away thinking after trying it.
Some 5-6 years ago I built a live streaming backend that used a custom C++ encoding front around Chromium on GPU servers on AWS. Mighty probably does something pretty similar. It was the right solution for our users at that moment because our product was a professional tool at a relatively high price tag, and being able to run full HTML+CSS allowed us to deliver customized motion graphics quickly using available designer talent, so there was a distinct advantage there.
It was also quite expensive to run, and the stack was fickle because those Nvidia GPU servers aren’t designed to run client apps — they’re AI/ML solutions primarily. So I’m not surprised that Mighty is shutting down: the infra bills must have been running pretty hot and they seem to believe they can better spend their cash runway on AI, presumably on largely the same infra.
(Btw I’m giving a talk at the Kranky Geek virtual event on Thursday Nov 17 about cloud compositing which will discuss this Chromium adventure and what I believe is a better solution for most apps.)
Mighty was like compute arbitrage. The bet was that you could buy compute at cloud prices, resell it to end users, and since this is cheaper than buying consumer compute in the form of workstations, laptops, & tablets, you can pocket the difference.
I can't think of a bigger slice of consumer compute than browser workload; so if the scheme doesn't work there, it won't work for anything. My conclusion is that compute arbitrage isn't viable for B2C. You will have to actually provide a service on top of the resale. For example Github Codespaces is reselling cloud compute while simultaneously solving infra-as-code pain points in the CI/CD pipeline.
I interviewed at a very successful startup last year that was basically doing the exact same thing as mighty. But they were selling it as primarily a security advantage (since its effectively an air gap computer) with additional advantages to performance.
The product itself was basically identical to Mighty. However, because it was marketed as a Security tool, they got customers from 9 of the 10 largest world banks, several government agencies, and so forth who bought licenses for every employee due to the security advantages that having an ephemeral server in the cloud provides. The companies enjoyed that there was also a performance benefit and they could skimp on physical workstations as a secondary benefit. But the product was selling licenses by the pallet load because of the security aspect, the performance was just a bonus.
Like I said, the product was effectively identical to what Mighty was doing. I think it was even younger than Mighty and was vastly more successful due to its market positioning.
Hasn't cloud compute always been far more expensive than local hardware? I don't see any arbitrage opportunity here.
Twenty years ago VDI wanted to replace desktop PCs with VMware ($$$) running on quad-socket servers ($$$) with fibre channel SAN storage ($$$). I didn't understand the economics then and I don't today.
Which means that if Mighty survived the M1 it would have been dead anyway (probably in not more than a few months) by Azure and AWS offering it once Mighty proved the market.
Not exactly, if the market was small it wouldn’t be worth AWS’s time to provide a turnkey. Plenty of small operations provide services backed by AWS without getting eaten for that reason.
I don't think it means services+arbitrage doesn't work. They were paying (I assume) cloud VM prices, and had to charge 20-30/month. If they had lower costs and were able to charge $5/month and still make money, we might not be having this conversation right now.
Services+Arbitrage definitely works. But Mighty didn't solve a painful-enough pain point.
The arbitrage was how Mighty made money (buy a cloud VM for $10/mo and sell it for $20 with a software wrapper).
But customers need a reason to pay for it. The advantages that Mighty offered weren't significant enough to impact most people. If they did, the advantages gained were questionable IMHO. Forcing people into a whole new workflow to slightly improve client-side JS performance is probably a worse trade-off for the vast majority of people even if you ignore cost entirely. Only a small group of people are encountering this pain point enough to actually seek out a solution for it. Once that small group of people find your solution, then you have to convince them to pay $20-30 a month to remove it. People are already canceling $12/mo Netflix subscriptions. To pay double-that for better JS rendering is a tough sale.
The most incredible thing about Mighty, for me, was that VCs bought into it. That... is just, wow.
I guess it goes to show that the idea is apparently only a small part of their decision making process - how it's sold to them is the bigger piece by far.
Mighty was conceived at a really dark time in Mac laptop performance.
The combination of Intel Mac Laptops that had horrible thermal characteristics, browsers that didn’t do anything to throttle background tabs, bloated websites and limited memory created a perfect storm of awful performance that Mighty could address.
Almost all of that isn’t true anymore (besides the bloated websites which browsers manage better), which decimated the potential Mighty market (people willing to spend hundreds a year on a fast browser).
I was a paid user, so I guess I'll chime in from that point of view. Why did I pay for this?
On my work computer, I keep everything separate. Chrome is work stuff. Mighty was my personal stuff, with little to no files stored locally. When I run all the docker containers to stand up the local server. My computer is very taxed for resources. Mighty was amazing and not bogging down my computer with a second browser open.
However, on my big beefy personal M1 Max MacBook Pro with 64 GB of ram. It actually felt slow. I found myself reaching for regular Chrome more and more.
Very niche use case. IMO they needed to pivot to B2B and not some niche B2C play, like what they were pursuing.
Overall, it worked impressively so. Kudos to the team for shipping something so stable that worked.
This idea sounded bad like a lot of ideas initially sound (insert ipod nomad lame meme). There was a case to be made that consumer laptop processing speeds had completely stagnated, chrome wasn't getting any lighter, and fast internet was more readily available.
But, Apple stepped in with the M1/M2 and totally obviated the need for any product like this.
The trick of Figma claims another. I see Mighty as part of the broader trend of folks thinking Figma it the leading edge of a wider transition to web apps across all creative work.
But all forms of coding and media editing (the main tasks that are truly CPU/GPU-bound, the area where something like Mighty would help) have remained as local desktop app affairs.
I think what really happened is folks misinterpreted the significance of Figma. The web has taken over in collaboration-focused software, e.g., things like Google Docs, Slack, etc... What really happened is that design has moved from specialized professional software (hard to use, powerful), to being more like collaboration software (easy to use, light).
I.e., what Figma really signals is that design is now more like new task that became more like using Google Docs app, not that high-powered tasks like video, photo, and audio editing are becoming web apps. This means what really powered Figma's takeover was that flat design took over, which is less technically demanding.
> But all forms of coding and media editing (the main tasks that are truly CPU/GPU-bound, the area where something like Mighty would help) have remained as local desktop app affairs.
Media creation maybe, but I feel like professional software engineering might move soon.
At the beginning of the year I moved from a company where I worked entirely locally, running a local stack, developing in a local editor, etc, to a company where I work almost entirely non-locally. That happens to be on a box under my desk, but I only use it over SSH. It's beefy, and I could get an even beefier VM somewhere if I wanted. All my editing is now in a web-based VSCode instance which has been much closer to desktop VSCode I was using before than I expected. All my builds happen remotely. It's honestly an amazing experience. I think things like GitHub Codespaces have so much potential here.
I agree overall, but I don't think it'll be VS Code in the browser, I think VS Code's remote features with the client-server model (client: VS Code's UI running locally, and server: the code executing and file-system running on the server).
I've also tried VS Code in the browser, and personally I find it an absolute unusable mess, the key binding space is just way to overloaded for a complex app like VS Code, and the browser itself to co-exist (e.g., many of VS Code's important bindings get eaten by the browser itself). I think that all that really matters is that the UI for complex apps runs locally.
(But I could be wrong here, I don't make the mistake of extrapolating my own experiences to other users. E.g., I also find VS Code to be so slow I avoid, but most users couldn't care less. https://news.ycombinator.com/item?id=27360494)
What hardware do you use to talk to the box under your desk? How do you work remotely with this setup, and what if the network VPN goes down?
Working in this way depends a lot on the stack you’re working with. For languages like Java or C#, where you can’t really work productively without an IDE, browser-based VSCode won’t cut it for most people. (I’d prefer a JetBrains IDE over VSCode too.) If your language doesn’t have a good remote debugging story, a remote-first stack won’t cut it.
This setup with a box under your desk doesn’t sound reasonable to me from a financial side either. According to your profile, you’re at Google, so I assume the hardware you use to access the box is a Chromebook. Sure, those are dirt cheap, but is a desktop + Chromebook combo really cheaper than one reasonably priced PC laptop with specs similar to the desktop?
I think I’m still happy with my beefy (spec-wise and kilogram-wise) laptop, having the ability to do everything directly on the machine (with no network round-trip for every operation), and still being able to do things if the VPN or the network goes down (with limitations, of course).
> What hardware do you use to talk to the box under your desk? How do you work remotely with this setup, and what if the network VPN goes down?
I use a MacBook Pro. I just connect to the machine. We don't use a VPN (see BeyondCorp) but something like Tailscale would probably work well. The fact the machine is under my desk and not in a data center is due to my laziness in returning it, nothing more.
I'm using VSCode for mostly Java. It's not perfect, but I prefer it to JetBrains products. JetBrains products are more capable for now, but I think their new IDE does remote development, and VSCode gets more IDE-like all the time. I think that's the direction things are going.
I'm using a MacBook Pro because I've been using Macs my whole life and have too much muscle memory to switch to something else.
As for having a beefy laptop, that only goes so far. Right now laptops tend to top-out at 32GB RAM if you still want plenty of choice, you make sacrifices to get more than that. Also many companies have policies or regulation restricting what can be on laptops.
If you are at home, your box is in the data center, then a BeyondCorp outage makes it impossible to work, right?
I agree that laptops can’t always do everything, but many things can be done comfortably well. At a previous job, I had 32 GB of physical RAM and 20 GB of swap (thanks to 5-10 RAM-hungry JVM apps), and things were quite usable for standard dev workflows.
Err, sort of. Beyond Corp isn't really one single service that can be running or down, it's more of an approach, a bunch of different pieces, etc. I haven't seen an outage yet, but I don't think it would look like "the VPN is down", probably more like "this web app thinks I'm logged out".
I do recommend reading the BeyondCorp paper, although it's been a long time since I read it.
VS code has the best remote features I have ever seen, good enough that I once got annoyed VS code had freezed just because the network died, then _realized I was doing the work over the network_. I hope it will stay a local app with powerful remote capabilities.
Figma is really slow on older computers. If your laptop is (say) 4 years old, it's entirely likely that Figma is not usable on it and so something like Mighty might make sense. But then when your power supply goes out and you replace the old laptop with even the cheapest Mac, suddenly you don't need Mighty anymore.
Trying again, there's one point: What really made Figma successful is design moving from being made in "professional software" (like say Photoshop, Premiere, Visual Studio, Maya) to "collaboration software" (like Slack, Google Sheets), and that this was made possible because "flat design" is less technically demanding than the "skeumorphic" design it replaced.
(Sorry I wasn't clearer, I struggle on how to make points like this succinctly without is sounding a bit disjointed.)
I agree with one of your points (that where Figma really succeeded is that it understood better than anyone else the collaborative nature of web design), but I disagree that flat design vs. skeuomorphic makes much of a difference. I've seen more "realistic" designs where the actual iconography/images are created in something like Photoshop and then just imported to Figma, and Figma handles it fine.
Point being that I think that even if all apps still had the design aesthetic of 2008 iPhone apps that Figma still would have succeeded.
I think the delta you're talking about is already gone. Figma is generally considered more performant with its WASM rendering approach than its desktop counterparts.
But I don't think the let's say "other deltas" that are now keeping other high-powered apps from becoming web apps are moving at all. You can check my other comments in this thread for more details, but a specific point I'm making with the original comment is that there are other areas the Figma hasn't dented at all that prevent more powerful apps from being adapted to the browser. And that the fact that design no longer needed those features is what paved the way for its success.
Figma as part of an app family, is way way more similar to Google Slides (i.e., office suite) than it is to Photoshop (i.e., "professional" software).
Multiplayer is such a game changer that all editing tools will eventually feature that, so the Figma model is going to win.
Once that is true, you want as many people to collaborate as possible and the browser is the perfect delivery model. Plus you ensure a consistent version across all clients.
I don't disagree with that point as an idea. But I do disagree that that's the direction the market is going. E.g., we're here discussing a startup pivoting away from a bet on exactly the idea you're expressing here.
Today the onus is on people who still believe in that idea to express why they think Mighty failed (without falling into the well-worn "year of the desktop linux" trap).
Personally I don't care about the idea either way, I'm just looking at is where the market is going (e.g., how the market share of various software packages are trending). Ideas that sound good fail all the time in the market.
And so far, outside of Figma, the idea you're expressing looks to me like a failure? And personally, I've started to look at other things different about Figma that might have accounted for its success.
> Figma as part of an app family, is way way more similar to Google Slides (i.e., office suite) than it is to Photoshop (i.e., "professional" software).
Digging into my own comment here, part of the reason I harp so much on the skeuomorphic to flat design change is it's such a bizarre thing to happen to an industry, for its defining requirements to change in such a way that the technology gets so much simpler.
E.g., can you imagine if realistic light rendering suddenly was no longer desirable in 3D software? Of course that would leave a gigantic opening for new players!
In the code space, would have to disagree I think, given how both JetBrains [1][2] and VS Code have embraced the client-server model, and how products like GitPod have used these to a full extent with excellent native integrations.
I think the distinction is that the UI needs to run locally so that shortcut-rich powerful apps don't have to fight for the keybinding space with the browser.
I'd say that the migration to the client-server approach for professional software is already underway, with VS Code of course being the canonical example, but my (limited) understanding is that Blackmagic's Da Vince Resolve (which is currently eating Premiere and Media Composer's lunch) also uses a similar model where data can be stored remotely but the UI runs locally.
> This means what really powered Figma's takeover was that flat design took over, which is less technically demanding.
Absolutely 100%. The other thing is that flat design brought a lot of low quality designers into the market because you could just add a few recommended spacing grids and flat colours and get a design out. But because the quality often was low, it forced collaboration for cross checking and that caused Figma and other software to come to the forefront.
According to the final tweet in that thread [0] it seems like they are pivoting to a sort of AI generated art based social media platform. I’m curious why they think this has the potential to be a “mass market changing” business since the social media world seems pretty over saturated already
He says in his 4th tweet in the thread "I think there’s a large opportunity to make an experience using many advances in AI (not just diffusion) to make a new kind of Creative Suite. A different set of products with a brand new UX." He links to the prototype at the end of the thread - https://playgroundai.com/
The current site doesn't look very inspiring. Like you say, just another "AI generated art based social media platform". But if his plans are to develop it into some sort of After Effects competitor then ... maybe that's a market opportunity worth cracking?
...but that particular industry is moving so fast it's already homogenized (with the best and cheapest tools being made by the core AI developers themselves), so without extreme differentiation a new player can't compete.
The beta app in the Twitter thread has less features than current open-source AI Image tooling.
In May 2021 Paul Graham wrote Crazy New Ideas[1], in which he explained that envy is the only reason people criticize ideas by "reasonable domain experts". This was widely understood at the time to refer to critics of Mighty[2][3].
I know to exist in the VC world you have to buy into insane shit and convince yourself you believe it is the next best thing, but this just seems so ludicrous to me. I don't understand it at all.
"the difficulty with finding truly transformative, great ideas is that they look pretty similar to bad ideas in the early stages" is one of pg's central theses that I couldn't disagree with more strongly, and the only conclusion I can come to is that pg's hubris is blinding him.
I don't think it's so much that you can't tell the difference between crazy bad and crazy good ideas in the early stages, it's that you're asking the wrong people. pg loves to quote the mythos that AirBnB was such a "crazy" idea early on that big VCs couldn't see the potential. But even in the early years, I don't think AirBnB was a big leap at all. Couchsurfing was already big, and platforms like HomeAway/VRBO had received huge funding rounds. pg was just asking the wrong VCs - not surprising that a bunch of rich old guys would be put off staying in someone's guestroom. Indeed, my understanding is that the first investor to really see value in AirBnB was very familiar with the vacation rental space.
In my opinion, the startups that were wildly successful weren't so much because they had "out there" ideas, but because their execution was unparalleled. Dropbox, Stripe, Figma, AirBnB, etc. just came out with products that were a delight to use when you first tried them in ways their competitors weren't.
Trying to be introspective, I think the event that most surprised me and seemed "crazy" was the election of Donald Trump in 2016. But again, I think i was just asking the wrong people. My circle of college-educated urbanites was basically unaware of the depth of some of the discontent with the status quo that Trump tapped into.
agreed, but on AirBnB, AirBnB was huge because it was cheaper than a hotel, they could ignore zoning laws, and it was meant to be for more personal & honest experiences. it's now far more expensive than a hotel, governments are restricting AirBnB subletting after complaints now, and the experiences are far more sterile with scams more common.
the complaints from the peanut gallery was always "this will work until it doesn't anymore", just like Uber, and that's exactly what's happening, it's crashing down now.
People with enough money sit and imagine all day tend to develop delusions of grandeur; see the various flavors of tribal wise men in history.
Not so bad in and of itself, it’s when individual’s delusions of self worth infect others and drag us along with.
Daily work should focus on tending to human biological needs and telling the delusional to mumble their gibberish in a corner aside from that.
I feel zero obligation to validate PG or Musk or the rest. Just people. Each one of billions. Their figurative identities as wealthy members of society is due to conformity to politically correct spoken tradition, not an indication they’re almighty.
We live in a world of stuff that was once viewed as insane shit, yet ended up to be the next best thing. Sometimes the stuff was insane shit at the time, the technology failed because it could not be ready at the time, yet we still ended up with something useful in the long run.
None of this is to say that all of those crazy ideas, or even any given crazy idea, will work out in the long run. The thing is, we cannot know until that time has come and gone.
That was my original reaction to Twitch, too. Who in their right mind would pay to watch someone else play a video game? It was ludicrous to me (and to some extent, it still is).
Mighty so obviously had no idea who its customer was. Can't afford good hardware to run CPU/GPU-intensive browser apps? Why not spend $420 a year on a browser. Is it for enterprise or consumer? Who knows. He spoke of it like a once in a generation company [1] and the fact they spent 3.5 years working a product that never even launched goes against what I thought his philosophy was.
Replit I feel suffers from the same problem and, possibly not coincidentally, PG talks about it the same way. Is it an IDE? Hosting platform? Education platform? How have they not found PMF after raising $100M over 8 years [2]? I feel like every time I see someone talk about the company on Twitter it's about them launching some brand new feature in a new area. On that, I also only ever see the same few people on Twitter chatting about it and have never encountered Replit in the wild. It's almost always Codepen, Codesandbox, JS Fiddle or Stackblitz.
It makes me seriously question PG's opinions on individual companies as his bias is so clear and it makes him appear so naive.
I was always quite skeptical of this being a popular use-case that grows over time.
Can the underlying technology be used to create a first class remote desktop experience? From the ones I've tried, I've found that NoMachine is among the best performing remote desktop tools. I never used Mighty but my guess is they must have achieved better performance than this? Was Mighty's technology tailored to Chrome only?
I’m actually happy that Mighty did not succeed. A more centralized web with dumb terminals is not what we need. I was always a bit disappointed to see Paul Graham hyping it up.
Suhail is super smart though so I’m sure their next chapter is going to be awesome!
Running a browser to connect to the cloud to run a browser to connect to the cloud to retrieve the contents of a single 2D page to recompress and send back to the original browser is now "the future of computing".
I think this piece from a few years ago about how difficult it is to build a startup around GPT-3 [1] is relevant to the AI art product they’re pivoting to. Best of luck to them, but the “build a product on top of open source AI models” space overwhelming favors the incumbents that actually are doing the cutting edge work in driving those models.
Good to see the pivot before cash ran out, but sad news :( Good luck to the team!
IMO Mighty is too early bc they didn't tackle the SW side, 'just' the remote nature, so couldn't get their 10X. The 10-100X shift today comes from apps running differently on cloud resources, not just lift-and-shift. That already exists as the VDI market.
Ex: We do client GPU <> cloud multi-GPU for cyber/fraud/etc analysts wanting to visually investigate how their many events stitch together, and Otoy does same for movie effects. In the consumer market, seeing same rewrites in say gaming. In a sense, Mighty's pivot to generative AI is the same -- faster to use remote multi-GPU services.. but custom built accelerated visual app for 100Xs, not lift-and-shift.
Without tackling the SW rewrite problem, hard for Mighty to get these wins via a lift-and-shift :( I do think federating cloud resources so users can bring their own and devs can reliably tap them would be amazing, but we seem still in the dev-controlled-server era. As someone trying to build predictable 100X experiences, that feels likely for awhile. Such a shift could have been webgl2 etc standards enabling modern multigpu apps in the browser... but Google and Apple web GPU browser/standards leadership have long strangled that path.
In a related note, we are starting to launch our global GPU edge network, and looking for a backend/infra eng on that + related AI services build out -- Nodejs/python/k8s. We are profitably growing: I agreed with Mighty's pattern before it even existed (my PhD at Berkeley explored it!), just easier ROI right now by sticking at the app layer.
I don’t think Mighty was unreasonable, just tackled from the wrong end. Netlify and Vercel did it from the other end and look where they are right now.
Instead of a faster browser they did super easy edge deployments of your JS apps which has a similar end result for users but also solves a giant headache for businesses which is where the money is.
It was very surprising to see how many people were buying into this (frankly awful) idea when it was posted here a year or two ago. Any one person criticizing it was immediately dogpiled on by 10 people telling them how wrong they were. Refreshing to see everything come full circle in the end.
In my opinion, on the face of it, remote experiences like this, be it gaming or whatever, are not out of the realm of reason. In practice, they tend to not work (again, in my opinion) because the rate at which computers get faster and cheaper far outpaces the infrastructure necessary to make a these "fly by wire" solutions work or have them be adopted.
If Moore's law (or it's equivalent in compute/dollar) really ever does level out, then this will become more attractive. With the cost of compute dropping exponentially, it doesn't seem like this can ever really get a foothold.
Seems like the privilege of raising a lot of money without having a product, revenue, or market fit can be a curse as well. They could have figured it sooner with less money in hand.
For those, like me who don't know what Mighty is: apparently a web browser that is faster. If that is the selling point, I can see why he quit working on it.
I don't think most people using browsers care how they work, so if that was the pitch, even worse.
This is from the home page, it just focuses on faster and more capable:
A new browser to work faster.
Mighty is a new browser that loads pages faster, finds docs quickly, and remains snappy with hundreds of tabs to save you time and make you more productive at work.
I actually thought about this company recently, because slack, zoom, Google suite, and all the other work apps perform so badly on my linux machine. Even with an i9 I get occasional full utilization. I thought I might just embrace the meme and outsource my browser. I only use chrome for work anyway. Too bad, Mighty, it is possible a similar business might work some day to centralize corporate work environments.
This is one startup I never "bought". Sure, they have an interesting product, but it is not something I would pay for
As other comment said, it's a vitamin, not a painkiller. And you can always buy a new computer or upgrade it. Would it really make me much difference to run a browser in a much more powerful computer than the one I'm at? Only in extremely limited situations
Suhail’s Twitter is definitely worth a follow as he is regularly tweeting about the everyday problems and challenges of running a startup. If you’re interested in the space it’s a good mix of technical and startup challenges and is one of the few Twitter accounts that provides great content all the time (in my opinion)
Although I follow him, I get such a feeling of arrogance from his tweets (similar to the Repl.it guy) that, although the failure of a startup is sad, I can't help feeling some satisfaction from this development (also partly from the vindication as I was certain that this wouldn't work as soon as it was announced).
It's decent tech, I don't think the market was there for it, though.
The problem with too many tabs wasn't primarily performance, although that was a part of it (disabling background tabs is a better optimization), it was that there were too many tabs.
So, cool tech but I don't think most people needed it.
It's basically Google Stadia for web browsers; do the computation and renderning in a far-away data center and stream the result as a video feed to a local thin client program. I suppose this is an idea that could work better for web browsers than video games, but modern web browsers and sites/"applications" require fewer resources than modern AAA video games by at least a tiny margin and the pricing just wasn't competitive.
Good pivot and a humbling reminder for us all. Jonathan Blow on the day this was announced, "Very hard to describe how embarrassing this is for everyone involved in the Web and, sort of, software more generally."
Hmm, I'm not quite sure how to interpret Jonathan Blow's tweet. You seem to be saying that he is saying that the Mighty project itself is embarrassing, while I took it to mean the fact that today's beefy computers, with all their CPU and RAM, struggle to render web pages to the point that someone is considering doing a rube-goldberg-esque solution of rendering on a mainframe-equivalent and then streaming you the results is the part that's embarrassing.
Yes, and from that perspective the fact that it didn't work and nobody actually wanted it seems to indicate that it shouldn't have been embarrassing after all.
With different people each time. And some of the people are on the correct side more often than others. It's useful to check afterward and update your priors on who you should be listening to.
Called it here back when it launched. Streaming video of a slightly more powerful browser in a browser (electron) never made sense to me.
And the few people who really had that usecase could just use a lightweight container running chromium + guacamole on a nearby server. Okay I'm sure Mighty used cool tech to make it faster than just that (idk how it worked but maybe they were transmitting changed HTML) but at its core the idea was not something normal people and 99% engineers would ever consider. Anyone rich enough to pay for it would just pay for a more powerful machine.
Anyways, they said they have 50% of VC money left and will be using it to create an online stable diffusion thingy which atleast fifteen startups are doing already including stability AI themselves. How can they differentiate ? Can they just use leftover VC money to make a completely different thing ?
I've just looked into Mighty. It has delays for hover css effects & scrolling. Also it has "Debug > Enable H264 encoding" option. So Mighty streams video.
I think replicating DOM & sending diffs instead of video should 1) saves resources on encoding for server 2) saves decoding on client 3) send much lower data 4) feels much better since instant scrolling/hover
(something like https://www.rrweb.io/)
As side effect of snapshots you can confidently unload documents & save CPU/memory.
Looking forward to grow that snapshotting part into DOM streaming. So you can run "browser server" on your own PC, while having laptop fast & cool. With almost 0 delays and without worrying about privacy.
From their description: We write code primarily in C++/JavaScript/TypeScript. Our application is a desktop Electron app where the front-end is in JS while the bits that need to be fast are written in C++.
Then we should make the opposite true as well, ie PG calling envy the only reason domain experts could be pessimistic about a future technology. He was too arrogant in those words such that when it failed, he should rightly be denigrated for his hubris.
Because each tweet is (possibly) worthy of separate discussion. you have
Tweet 1: thesis.
t2: supporting evidence A.
t3: supporting evidence B.
Say you agree with the thesis but think evidence A is weak or otherwise want to discuss it, you don't pollute the discussion on the thesis or supporting evidence B.
Mighty as an idea was cool and interesting and I hope the tech survives and continues to be developed. Fat clients have their place in the world but a solution like Mighty allows for a lot of the benefits of a fat client without a lot of the downsides.
I think this is a really interesting decision from Suhail and one not taken lightly. I think this decision is another data point that AI-infused applications are a potential new “tech platform” and we’re at the beginning of a new “mega cycle”. I.e. web2 2002-2010, mobile 2010-2020.
Folks shouldn’t be spending much energy on thinking about how Mighty may have gone wrong (and seriously the negative analysis is so boring and lame), and instead think about the new AI opportunity.
Before the value proposition was good. You can still use your old laptop and still have a very fast browser experience via remotely running it on a AMD server chip, this way you save on battery life and RAM too.
Then M1 speedometer scores started beating AMD Epyc on single core web browsing performance. Not only that, but even providing a much better experience in terms of battery life because of power savings.
Then yeah there was no point paying 20$-30$ a month for a remote browser instance. If you can afford that as a user you're either already set on CPU/GPU/RAM or better off just buying a used m1 air for <900$.
It's quite sad of a lesson where you do everything right, work consistently hard, be quite innovative, have plenty of financial and talent resources, and still fail.
Tech obsolescence is brutal.