Comment by GuiA | Hacker News Reader

Comment by GuiA | Hacker News Reader

I don't want more shit relying on a 1 GBps, sub millisecond latency internet connection. Even if bandwidth has gotten better in the past decade (although it's doing so logarithmically), latency is still there no matter what you do, and sadly the speed of light makes it that no matter what you do, you can't do NYC-Paris in less than 40 milliseconds. You can use tons of tricks to somewhat massage that fact, but at the end of the day the physical limit is there.

Not everyone is living in your Google Fiber bubble. If you're going to design the cellphone of the future, you probably don't want the base assumption of your design to be that the user will always be in an area with high bandwidth, low latency environment, because that represents a minuscule fraction of cellphone users.

Sadly, that's what happens when all of the startups are based in an echo chamber - they lose sight of the fact that not everyone has the same conditions as them. Well, tech has the nice habit of cyclically reinventing shit, so I'm glad we're going to spend a few years banging our heads against thin clients again.

Homework reading for the founders and employees of this company: https://rescomp.stanford.edu/~cheshire/rants/Latency.html

The odd thing to me is that these guys live in San Francisco (I met most of them a few years ago). Wi-Fi is not ubiquitous here, Google Fiber will never be deployed here and most of the city only has Comcast as a real option. Their service is as good as their reputation would lead you to expect. Nextbit should know first hand that the world is not a well connected mesh of low latency, high bandwidth, and quota free connections.

The idea of loading what the phone thinks you'll need into its local storage is intriguing, and people keep trying to make it work to varying degrees. I don't think it has ever, or can ever, work simply because of the massive constraints it puts on usage. Even if the phone is right 99% of the time, the first time I need to use Google Maps and its been "uploaded" off of my phone, it's going in the trash. Just the thought that such a thing might happen and be beyond my control makes me disinclined to give it a shot (not that I would ever own an Android phone anyway, so no sale lost).

I think that devices that try to anticipate the user's wishes and plan ahead, but which may make decisions that have negative consequences, violate the user's trust. I expect technology to do what I have asked it to do -- in this case, I asked it to store something and it thought better of that.

I don't really understand their cloud storage model either. 100GB on S3 for 3 years is $99 -- not counting bandwidth and EC2 instances, which is what really racks up the bill. People aren't going to use their entire allotment all the time, but this ongoing cost is going to sink the company.

You average user doesn't have close to 100GB to store and when you dedupe the big media files that drops even more. Hell, Amazon is doing unlimited storage for consumers for $70/year already and that's not just for phones.

I agree that most users don't utilize all of the space provided by any online service, although in this case what they upload is also out of their control. Typically space utilization is somewhat hampered because users won't be bothered to wring every last bit out of it by hand.

Even if we say each user only stores 10GB over 3 years, that's $10 just in S3 storage costs. Bandwidth costs for S3 can easily eclipse that if data is transferring back out to the device frequently. Even at a super conservative 10GB outbound per year, that's another $1.50.

I'm skeptical that a $400 phone sold in small quantities has the kind of margin to support this type of ongoing cost over the long term. They're smart people though, so maybe they have factored all of this in.

Your average user doesn't have close to 100GB, but your average user also has no use for an infinite storage service like this. So their average user will not be the average user.

Selling it as infinite is great for people like us that want that, but is also one less thing to worry about for the gut who doesn't need that much space. Infinite is better than even 100TB to the average Joe.

Physical limits can be worked around with tricks though. For example google had a feature where they prefetched and started loading a result as soon as you start moving your mouse towards it to click. Latency is still there but hidden from the user.

Pretty excited to see what innovative tricks with predictive prefetching they came up with to base a whole company on it. I can also see this being useful for simply just files instead of apps. For example Dropbox on mobile also has a "thin client" approach but doesn't do anything smart to cache frequently accessed files unless you explicitly favorite them.

> For example google had a feature where they prefetched and started loading a result as soon as you start moving your mouse towards it to click.

This isn't a trick, it's moving the information nearer to your user. This comes with a maintenance cost (though cache patterns make this much easier).

The size of the files here are big enough that bandwidth is more important than latency. And by default it will only happen when the phone is plugged in and connected to wifi.

I'm basing my criticism of their vision based on statements like this one:

"By connecting your phone to the Internet—not just through apps and notifications, but at every corner of the operating system [...]"

And mistaken assumptions such as this one:

“When iOS started development and Android started development, that was 10 years ago. Networks were much slower, Wi-Fi was much less ubiquitous, and AWS hadn’t democratized cloud to the extent that it has, right? So because of that, a lot of choices were made that result in user pain points today, which could actually be alleviated by leveraging the cloud in the way we’re doing.”

Guess what, WiFi still isn't ubiquitous for 90%+ of the population that uses smartphones, and probably won't be anytime soon (think 3rd world countries, rural areas, etc.)

I live in a large city and I still don't always have wifi. I have wifi at my home, obviously, but not on my daily commute, not at most grocery stores, not when I'm out hiking in a park. During those times something like this would just be sucking up my limited lte bandwidth. Heck I'm sure most people still experience losing their cell signal when in the guts of a large building. If I really need more space an SD card seems like a much better option.

I may be in the minority here, but even if it was ubiquitous I wouldn't want "every corner of the operating system" running through public WiFi.

The creators must be aware of these concerns. I'd be interested in hearing their thoughts on the issue.

I kinda really hate how City wifi is being done. Every time you connect to the network you have to go through the damn login process on a web page because obviously TWC can't have any of those nasty Verizon customers logging on. I don't even bother with it anymore because all it does is block whatever it was I was looking at as soon as I step out of my house.

It doesn't have to be ubiquitous. The only times you're going to run out of space is when you're downloading something, and chances are good that you'll run out when you're downloading something large. So you'll already be in a position to download a large file when this kicks in.

Another use case for running out of space: recording video. Which is quite likely to be when you're out and about someplace with no WiFi, like a park, for example.

Most of all: the only real reason to force me to store my stuff in your cloud is so you can fossick through my stuff to make more money out of me.

Reply on news.ycombinator.com