Comment by vishnugupta | Hacker News Reader

Comment by vishnugupta | Hacker News Reader

Parent Full thread

There was this product[1] to browse internet offline by downloading something called “Web Packs”. This was back in 2005 just when I was graduating when I spoke to the founding team. They were naturally quite confident about the product taking off. Something seemed off to me I couldn’t point out what so I didn’t take the job offer. After all these years I realized the source my discomfort. They were actually betting against the speed of internet getting better. While most of the businesses like Amazon, Google were betting for the internet tech to improve this product did exactly the opposite.

To me Mighty sounds like a similar category of product. They are betting that PC/Laptop/Mobile hardware will stagnate from this point on. Exactly when Apple has launched M1, which blows the previous version out of the water, at a non crazy price. From this point it’s a matter of time other hardware also catches up in terms of performance and price.

Besides, yet another company to handover my entire browsing history and data for purported improvement in latency? I don’t know.

[1] https://en.wikipedia.org/wiki/Webaroo

> They are betting that PC/Laptop/Mobile hardware will stagnate from this point on

No, they are betting that no matter the processing power of computing devices of the day, developers will always push things to the limit and ship products that can end up being slow on devices used by a large share of the population.

This has always happened, because software product makers always want their products to have as many features/capabilities as possible, and release them to market as quickly/cheaply as possible, which means there are lesser incentives to limit functionality or to invest in performance optimisation.

There's no reason to believe this trend will cease, so Mighty is offering a service that allows people to get much higher performance of their webapps without always needing to have the highest-power computing device in their possession.

> developers will always push things to the limit

Agree, it's an arms race at this point between end consumers and app developers. However, it also means once Mighty becomes popular they have to contend with developers overwhelming their computing resources as well.

As long as Mighty keeps providing a better experience than the local browser on the average user's own device, they'll have utility for a significant number of people. Developers still need to provide satisfactory-enough performance to the bulk of their audience. Mighty is for the users in the right tail who are willing to pay for better-than-satisfactory. There's no reason to believe that right tail will ever disappear.

Yes and mighty will keep increasing prices as the developers can now ship any horribly optimized thing.

Mighty can succeed yes, but it's success would be bad for the ecosystem.

Centralized processing of all computing, would just make current issues like censorship, security issues worse.

> developers can now ship any horribly optimized thing

No, they can't. Only the right-tail power users will ever pay for this. Unless almost all of a product's users are in that group, the developers still need to offer satisfactory performance to the rest of their users, or be vulnerable to a new competitor.

> Centralized processing of all computing, would just make current issues like censorship, security issues worse.

People keep saying this. I doubt even Mighty believes they or their paradigm will take over "all computing" - just the niche that really needs it. Even a small market share can make them a big/successful company, but without making the whole internet significantly more centralised than it already is. I don't get the panic.

The Mighty model is timesharing CPUs. At least 80% of the time my computer isn't even used. And when I use it, I doubt average load is over 25%. So the Mighty model has a 20x utilization advantage by those numbers.

That should easily support twice the peak CPU power for me when I need it.

Another thing is that the Mighty CPUs will (presumably) be upgraded continuously, while my laptop CPU gets no faster after purchase. If that makes me only buy a new laptop every 4 years instead of 2, I've saved a lot of money and hassle.

I'm not saying this means Mighty will conquer the world. But there are reasonable arguments behind the model. Especially if you assume bandwidth will keep improving.

>The Mighty model is timesharing CPUs.

It's not. It's literally spinning up a VM in the cloud to run Chrome and stream a video to you

> At least 80% of the time my computer isn't even used.

Yes, but Mighty isn't running (and will never run) on your computer.

Wow you are really not understanding the comment you are replying to. He's discussing the business model.

And I've responded with the reality of the business model. There's no "CPU sharing" on "my computer". It's a beefy VM in the cloud.

Well, it most likely shares the CPU with other VMs, but that depends on the cloud, and the instance. And since the browser is always open, Mighty will always run that VM, with no sharing: https://news.ycombinator.com/item?id=27063554

>> And when I use it, I doubt average load is over 25%.

So the idea is you can get your laptop utilization down to 1% by pushing your load to their cloud? Fascinating.

Hmm, I feel like this is missing that those cpu workloads are "bursty", and probably all burst around the same time for a given region. This analysis of unused computer time assumes they can sell your unused time to someone else, which either means network latency to another region or more capacity in the same region as you when you have similar usage times as everyone else. I have no idea if this idea works but I don't think it does for that reason.

> The Mighty model is timesharing CPUs.

Yeah, but I think it's better to think about it like virtual/remote desktops, where the granularity is a browser tab.

At least, that's the only way it makes sense to me. I might understand that to do 4k high res 3d on my phone - my phone will need help. And once that "help" is available - my TV, and my tablet and my smart watch can make use of it.

But I'm not generally willing to trade latency for cpu time sharing. What is interesting is an always on, always working desktop session. It's why I like screen/tmux and ssh, rdp - and would consider running a Linux terminal server, so my laptop(s) and desktop(s) could be a disk less, stateless thin client.

Make the observation that the browser tab is the new process/application granuality - and it makes sense to host tabs in the cloud.

Personally I'd want to self host it - but the idea doesn't sound quite so inane.

"They are betting that PC/Laptop/Mobile hardware will stagnate from this point on."

Me playing the devils advocate...

Perhaps not //forever//, but perhaps for the next few years (enough for them to make some good money).

Why? Chip shortage, GPU shortage.

Anecdotally, I just bought a computer for $7500, that, two years ago would have cost about $2500 (granted, it is the new tech, but even so, new tech of this tier two years ago would have been approx. $2500). On top of that, have to wait two months for it to be assembled. (ouch)

I agree with you that speed will probably not stop increasing, but if prices continue to go up, or even just level off, few people will be able to afford the new tech.

That being said, would I invest in Mighty? No since I agree with you in general....and agree with you about the zero privacy issue...very regressive position for someone like PG imo.

That assumes that AWS/Azure/GCP won't have the same issues getting chips to run Mighty on that PC manufacturers have getting chips to sell to consumers though. If the shortage keeps up, that is not necessarily true.

What if the future is that you don't necessarily just buy a machine in the future.

What if, when you purchase your machine, it includes a host of features including a could suite of things like; VPN, Cloud Browser, An actual amount of decent cloud storage, portable applications, hosted in your cloud bucket, but accessible from any device (If I have Illustrator, I can just run it from any machine - not just my own. I can allow guest access to my paid licenses - like Say i want my brother to be able to draw stuff on my Illustrator license while I am not using it.

Etc.

Imagine if instead of that $7,500 "computer" you bought a $7,500 'stack' and the access terminal you happen to be using most is the physical laptop that youre used to.

I tried to build something similar to this more than a decade ago (2007 or so - but wrote about it in 2004).

My biggest issue is better information/knowledge management.

I have TBs of files and data strewn all over. We should be focusing on "my information" and let me manage and secure that - and access it from any device easily, and securely.

If augmented reality tech works better the necessity of cloud as your machine will become ubiquitous.

However I think we gonna do have RTX in our mobile devices - AR glasses, whatever modern mobile tech - because of decentralization (security/ethics purpose)

Couldn't they instead be betting that despite improvements in hardware speed the web will bloat faster than the hardware improves? Betting on that doesn't sound that crazy to me.

That is indeed one of their premise. However that also has, perhaps unforeseen, implications. To start with, their VMs have to be at least one step better than the end consumer's machine. Can they do it without increasing their price? Secondly, there's no guarantee that developers will overwhelm even Mighty's cloud resources?

All that said, it's worth noting that the first set of customers seem super impressed. So maybe Mighty are on to something.

Not only that but you can just use a different browser.

Reply on news.ycombinator.com