Comment by vitus | Hacker News Reader

Comment by vitus | Hacker News Reader

Parent Full thread

5 Mbps down is certainly more than adequate for a single 1080p stream that's been compressed ahead of time. When you have to stitch together multiple real-time streams as in a classroom environment, then the actual numbers can vary significantly.

If your kids are forced into remote learning, and you're working from home, then we might be talking more like 3-4 Mbps per person if your network is split 5 or 6 ways. And unless you're on a business internet plan, the only guarantee you have when you buy a 20Mbps plan is that your ISP will throttle you to that rate if you exceed it.

To be clear: I think most people would be happy if they could get a guarantee of 50 Mbps download speeds. But the fact of the matter is that during peak hours, if you pay for higher download speeds, you get a higher fraction of the available spectrum. Your big-name cable companies can have oversubscription rates of 100x or more. That doesn't translate to only getting 5Mbps out of your 500Mbps connection at peak, but you will see _some_ contention. If you're paying for 20Mbps under the assumption that you'll only need 18Mbps at peak, you'll be frustrated when you end up only getting 15Mbps and all the video calls stutter / cut out intermittently.

> If you're paying for 20Mbps under the assumption that you'll only need 18Mbps at peak, you'll be frustrated when you end up only getting 15Mbps and all the video calls stutter / cut out intermittently.

But isn't that the real problem here? Companies shouldn't be able to advertise a 20 mbps connection if they aren't able to maintain that speed. Dropping one megabit on occasion may be reasonable, but five is at the point where you should be compensating customers for the outage.

My thesis is not "everything is fine and everyone should shut up and accept what they've got." My thesis is that we're focusing on entirely the wrong thing.

There are certainly a lot of problems with incumbent ISPs that many other commenters have pointed out: deceptive marketing practices, anti-competitive behavior (ranging from lobby-driven legislation to lock out new entrants, to outright sabotage / destruction of cables), wildly asymmetric up/down speeds driven by economies of scale from 20 years ago, just to name a few.

Yes, even 4k video streaming at 20 Mbps is viable (assuming that your connection is stable). Yes, poor wifi performance / latency / bufferbloat / squirrels eating your coax are more common underlying causes of end-user pain points. But that doesn't mean that there aren't common use cases that benefit significantly from adding more bandwidth, nor does it mean that adding more upstream capacity is useless in the context of these other factors.

Also, let's be realistic: most regulators don't have anywhere close to the level of technical expertise that you'd expect from, say, a network engineer. It's a lot easier to reduce everything to the single dimension that is download speed, and it's a lot easier for a layperson to understand said changes.

It's also a straightforward way to place pressure on ISPs to upgrade their networks if they want to keep saying that X% of their customers have access to broadband plans.

> But that doesn't mean that there aren't common use cases that benefit significantly from adding more bandwidth, nor does it mean that adding more upstream capacity is useless in the context of these other factors.

I agree! I just don't think it's what we should be pushing for right now.

> Also, let's be realistic: most regulators don't have anywhere close to the level of technical expertise that you'd expect from, say, a network engine

Well, I have higher standards than you do. I think regulators should have a very strong understanding of what they're regulating, or should talk to people who do.

> It's a lot easier to reduce everything to the single dimension that is download speed, and it's a lot easier for a layperson to understand said changes.

To me, this is a bit like saying we should still be measuring all CPUs according to clock-speed, because it's easier for a layperson to understand a single number. (And make no mistake, clock-speed is important!)

People here seem to think I'm letting the ISPs off the hook somehow. On the contrary, I think this fuss over bandwidth is great for the ISPs. People pay more for faster speeds because they think it's making their Zoom calls more reliable, when the actual culprit is just the piece of crap wifi router they're renting for $5 per month...

> I agree! I just don't think it's what we should be pushing for right now.

While I agree that it's not the only problem at hand (and perhaps not the single most useful change the FCC could make), I think any movement is better than none. If nothing else, the WFH / distance learning situation brought about by the pandemic demonstrated that the current standard for upload speeds is entirely inadequate (and so the proposal increases that from 3 Mbps to 20).

It's also possible that these government agencies are nervous about trying entirely new things, lest the current Supreme Court swat down any such efforts as perceived overreach (as they've done recently to the EPA).

> I think regulators should have a very strong understanding of what they're regulating, or should talk to people who do.

Should != do

It certainly would be nice, but in the last 20 years, the only FCC chairs with relevant prior industry experience were Ajit Pai (Verizon counsel) and Tom Wheeler (CEO of CTIA).

As far as talking to people with experience, guess who's most likely to make sure they're represented at the table? That's right, the incumbent ISPs and their lobbyists.

> To me, this is a bit like saying we should still be measuring all CPUs according to clock-speed, because it's easier for a layperson to understand a single number. (And make no mistake, clock-speed is important!)

Well, we sort of still are, except that one number is Geekbench or whichever other benchmark is popular these days. (Okay, there's also single-threaded vs multi-threaded perf, but there's a strong correlation between the two for consumer-grade CPUs.)

(and I'd expect that your typical consumer doesn't really care much about the CPU details in large part because performance is so opaque now. They probably just buy something with an i7 or the latest generation Macbook and call it a day.)

Are you literally incapable of understanding that other peoples' needs are different from your own?

Reply on news.ycombinator.com