I decided to stop working on Mighty(twitter.com) |
I decided to stop working on Mighty(twitter.com) |
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.
https://twitter.com/Jonathan_Blow/status/1387094702139142145
> Wait, Paul Graham supported this?
Jonathan: Very vocally, to the point of insulting world-class programmers who think it’s dumb (he did not know said programmers were world-class).
> Huh, I see. I thought he's all pro elegant and lean solutions, judging by the books and articles at least.
Jonathan: Yeah it doesn't make sense to me either.
https://twitter.com/Jonathan_Blow/status/1387121451187118080
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 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.
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/)
I'm indie/solo making Linkkraft browser (to make a living from it). Browser to be effective researcher & collector. It visualizes your steps as tree and makes html snapshot for your each step (even steps in SPAs like twitter). https://arestov.github.io/linkkraft-notes/comparing/linkkraf... https://arestov.github.io/linkkraft-notes/trails-tree-plus-o...
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.
PS. Linkkraft looks cool
Wait, the client was an Electron app? Are you serious?
If so, this just goes from a poorly-executed idea to an outright fraud.
After significant searching, I found it:
Basically Stadia for webpages.
Always seemed like a very, very niche product .
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".
[1] https://www.allencheng.com/starting-a-business-around-gpt-3-...
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.
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.
So much interest and explosion.
I’m tired of all these apps constantly doing stuff in the background and tracking my location anytime I use them.
If I had a “mighty” phone app, and could stream an emulated device with the apps on that server I’d happily pay $5~$10/ month.
If Moore's law (or it's equivalent in compute/dollar) really ever does level out, then this will become more attractive. With the cost of compute dropping exponentially, it doesn't seem like this can ever really get a foothold.
That was the selling point.
This is from the home page, it just focuses on faster and more capable:
A new browser to work faster. Mighty is a new browser that loads pages faster, finds docs quickly, and remains snappy with hundreds of tabs to save you time and make you more productive at work.
As other comment said, it's a vitamin, not a painkiller. And you can always buy a new computer or upgrade it. Would it really make me much difference to run a browser in a much more powerful computer than the one I'm at? Only in extremely limited situations
The problem with too many tabs wasn't primarily performance, although that was a part of it (disabling background tabs is a better optimization), it was that there were too many tabs.
So, cool tech but I don't think most people needed it.
I feel my web experience is mostly bound by other browser things than JS. I wonder what would be the best way to profile this.
Before the value proposition was good. You can still use your old laptop and still have a very fast browser experience via remotely running it on a AMD server chip, this way you save on battery life and RAM too.
Then M1 speedometer scores started beating AMD Epyc on single core web browsing performance. Not only that, but even providing a much better experience in terms of battery life because of power savings.
Then yeah there was no point paying 20$-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.
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.
[1] https://twitter.com/stephsmithio/status/1501729837366452224
[2] https://twitter.com/ID_AA_Carmack/status/1590816470845771777
> where you do everything right
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 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.
That price point was a serious barrier to entry.
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.
> M1
To have a M1 chip, you need a new laptop. A new Intel/AMD laptop would bring you plenty performance not to require remote desktop as a service.
Sorry what? There are people that even considered this?
It’s similar to people in jobs that require them to live out of their inbox paying (a comparable amount) for Superhuman.
Reselling Hetzner instances and locking access to Chromium, brought to you by the finest VC backed startups.
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?
Xbox Cloud, Playstation, Nvidia Geforce, Shadow (even has streaming VR), Amazon Luna, Paperspace, Boosteroid, etc etc. There's no shortage of options.
Google shutting down a business isn't a good indicator of a market's validity.
The "Buy why?" was simple: it made the internet very fast and improved battery life of laptops.
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.
Probably high server costs and being a vitamin and not a painkiller killed it.
Also, people are not too happy about giving away their browser history data to server-side powered browsers.
That being said, it's still sad to see startups fail. Hopefully they'll have better luck with their new direction. Fingers crossed!
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.
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.
I think it was already dead at the time and the criticism got under his skin.
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.
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!
[1] https://twitter.com/Suhail/status/1588906086459150337?s=20&t...
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.
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.
The render, compress, send over IP, decompress pipeline will just always feel laggy and slow.
> We're excited to finally unveil Mighty, a faster browser that is entirely streamed from a powerful computer in the cloud.
Source: https://blog.mightyapp.com/mightys-secret-plan-to-invent-the...
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.)
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.
Doomed for failure from the start.
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.
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).
https://addons.mozilla.org/en-US/firefox/addon/auto-tab-disc...
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.
But, Apple stepped in with the M1/M2 and totally obviated the need for any product like this.
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.
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?
PMF? What's that?
...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.
The social aspect was just added as an extra feature
It's more like a Photoshop + canva type project
[1]: http://paulgraham.com/newideas.html
[2]: https://news.ycombinator.com/item?id=27062713
[3]: https://www.echevarria.io/blog/the-mighty-pushback-isnt-all-...
https://twitter.com/paulg/status/1357097710734749700?lang=en
I dig when billionaires are wrong. They're just like us. :)
Mighty https://twitter.com/search?q=mighty%20(from%3Apaulg)&src=typ...
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 - https://twitter.com/search?q=replit%20(from%3Apaulg)&src=typ...
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.
[1] https://twitter.com/paulg/status/1561473124389888000
[2] https://www.crunchbase.com/organization/replit/company_finan...
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?
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?
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.
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..
To the parent post, there’s already lots of solutions for non performance oriented Remote Desktop environments.
Not really, the client can still OCR whatever the browser viewport contains, even if the server just streams video.
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.
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.
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.
> 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!
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.
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...
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
Heroin is.
Being a project with no customer base killed it.
I remember it being discussed 6 months or a year ago on HN and while it was a popular topic, it was not being discussed kindly.
It's something that should have never made it past the stage of cool demo.
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.
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."
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.
That's a great idea for a startup, VCs would be falling over themselves for a piece!
(Though I bet your won’t do it as often as you would have thought, now that your money is on the line. Pun intended.)
The only utility is vengefulness, perhaps.
Then you have to choose/agree and pay some kind of arbitration.
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)
But the trick, in this case is parsimony, the opposite strategy of what VCs do.
Only shorting when I am 100% positive the idea is a fail.
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!
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.
Just thinking of the literal mountains of ewaste gives me goosebumps.
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.
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.
It is our job to care so non IT people don't have to.
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.
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”?
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.
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.
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.
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.
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.
The success of AWS and Azure show it can be done.
(You do have to avoid replacing cheap commodity desktops with exotic bleeding edge servers, it’s hard to make up for that)
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.
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'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)
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).
This is not about web vs. native but client-side vs. server-side rendering & compute.
(Sorry I wasn't clearer, I struggle on how to make points like this succinctly without is sounding a bit disjointed.)
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).
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.
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.
It is so seamless that you forget that the code is not in your machine.
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.
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.
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.
I'd make a terrible VC.
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
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.
[1] http://www.chem.gla.ac.uk/cronin/news/cronin-group-builds-ch...
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.
Uber had to fight an absolutely massive taxi lobbies and what they were doing was illegal in a ton of cities.
With Airbnb, you had hotel lobbies and the idea of letting a stranger come stay in your home didn't seem like it would become a mainstream thing.
Anyway, you've been able to easily short crypto for the last 8 years. Have you shorted any at all?
FWIW, I'm incredibly picky about shortcuts, and as long as you use Chrome to "install" the web app, it basically works 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.
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.
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.
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
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.
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.
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.
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.
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.
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.
And liveblocks.io is here to make that happen at scale.
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!
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.
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.
Once you start having a meta discussion about the analogy itself, perhaps that should be a sign.
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.
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.
Yes. So does a VPN outage if you're using that instead. If your laptop gets chewed up by a lawnmower, it's also impossible to work.
What can I do without letting them in my actual browser? These people are crazy.
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.)
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 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.
I think baas is bullshit for security reasons alone.
Some people do all their work in browsers.
[1] https://en.wikipedia.org/wiki/Obelus#:~:text=It%20also%20has....
The people saying this either haven't used it recently or need a modern ISP
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.
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.
https://www.statista.com/statistics/999383/total-number-of-g...
Cloud gaming has a very bright future. If not being the future.
Especially with regards to something like input lag.
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).
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.
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.
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.
[1] (2017) Average U.S. household now has 7 screens, report finds https://www.fiercevideo.com/cable/average-u-s-household-now-...
[2] (2019) U.S. Households Have an Average of 11 Connected Devices — And 5G Should Push That Even Higher https://variety.com/2019/digital/news/u-s-households-have-an...
[3] https://www.rts.com/resources/guides/food-waste-america
[4] https://www.theworldcounts.com/challenges/people-and-poverty...
With 123 million households in the US, the mean spending on laptops, desktops and tablets together is $350/year per household.
Smartphone sales were $74 billion [4] or $600/household/year.
That's a total of $950/year, per household.
The average American household isn't spending $11,000+ every 2 years.
[1] https://www.statista.com/outlook/cmo/consumer-electronics/co... [2] https://www.statista.com/statistics/722371/united-states-des... [3] https://www.statista.com/outlook/cmo/consumer-electronics/co... [4] https://www.statista.com/statistics/191985/sales-of-smartpho...
"Apple generated $365 billion revenue in 2021, 52% came from iPhone sales" [1] Taking Apple as example, those billions have to come from someone. The greater point was that instead of having the billions try to make every single device the most powerful CPU/GPU of the current Moore's Law peak, because no one wants a slow device, you might just as well have a bare glass able to render pixels computed on a remote powerful machine, but that machine must be owned by the user. And you don't stream an app, a browser, you stream an entire OS, on the spot, on request.
My oldest (working) devices are already all closing in on or past a decade. This metric is flawed in many ways...
[1] https://www.cnet.com/culture/study-u-s-adults-spend-1200-a-y...
[2] The average US family size is 2.51 [3] while average worldwide is 4.9 [4]
[3] https://www.statista.com/statistics/183648/average-size-of-h...
[3] https://www.pewresearch.org/fact-tank/2020/03/31/with-billio...
Most people make a 1k$ purchase last decades. Enthusiasts may be dropping 5-10k$ but not your average joe.
In 2005 the average size of a TV was 30 inches. Two decades ago a TV had no HDMI, and it only received analog NTSC or PAL.
A phone was 2G - 3G was the thing about to be but mostly not there yet. One decade ago the iPhone 5 was released.
An Apple iBook cost $1200 with 128MB of ram and a PPC processor: https://apple-history.com/ibook_mid_2002
Maybe a stereo is usable - although for midrange equipment the capacitors have probably dried out and you certainly won't be getting digital FM.
You would need to be very poor if you can't get working hand-me-downs from a decade ago.
People do it with cars (even if they're not leasing, which is explicitly doing it).
I can think of quite a few electronics I spent money this year which did not include any screen device. Again, these stats are misleading, maybe dishonest.
My "device" budget we'll call it has been less than 1000$ 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.
As a general heuristic, one single person's decision tree is rather irrelevant in the long open game of average spending behaviours.
Ha, as if.
"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.