[bawolff]

> My bet is always-on broadband. Virtual worlds, digital entertainment, gaming …. None of it would be possible without broadband. Today, we have about 100-250 Mbps in most modern homes. In some places, you can get a gigabit per second. Now imagine what we could do if, in ten years, we all have 10 gigabits per second in our homes, and we have in-home networks that are fast and fat.

Hmm. Kind of an interesting take. I predict the opposite. I think we have hit the diminishing returns on bandwidth. People don't even know what to do with residential internet in the 100mbps. We can stream HD video (both up and down)...what else is there. If there are no applications straining at bandwidth limits now, im not sure improving them will make new products. If anything i think the best gains are going to be in latency and in mobile bandwidth. Cheap high speed, no data cap, internet for cell phones would be pretty great. But still i'd consider that incremental. I think its unlikely the next big technology enabler will be the same as the last one.

> In 2001, we imagined a 100 Mbps future — and we got Google, a nearly trillion-dollar company.

Google was founded in 1998, and that was still the dial-up era. At the time, I was probably thinking a 1mbps future sounded pretty grand and couldn't imagine 100mbps. Heck my internet right now isn't even that fast.

[jordanthoms]

I'd settle for IP-based video conferencing that actually works, 100% of the time. As far as I can tell the major issues with that now are more from WiFi - Bufferbloat is particularly bad there and interference causes small dropouts (run a ping to a local router in the background for a while, and you'll probably see jitter eventually).

It's pretty noticeable here in New Zealand where most people are now on fiber connections with easily enough bandwidth and low latency - but video calling is still way harder than it should be.

[yomly]

Are you talking about video chat within NZ or with rest of world? I imagine there's a fixed amount of latency with calling anyone in NZ due to speed of light...

[fastball]

As the crow flies, it takes light 66ms to get from one point on earth to a point on the opposite side.

Pretty sure the limiting factor is not the speed of light. It's the fact that (currently) these packets on not traveling in a straight line, and there is a lot of switching happening between A and B.

There is plenty of room for improvement in this area.

[jack1243star]

> IP-based video conferencing that actually works, 100% of the time.

This is oxymoron. Our packet-switched infra simply cannot fulfill such promise.

[bawolff]

That's a bit pedantic - the poster would cleary be happy with something that worked 99.9% of the time, and there is no technical reason why our packet switched network can't suppot that in theory

[EdJiang]

For all the net-neutrality proponents: isn't this an argument against net neutrality? If ISPs could discriminate depending on traffic and apply better QoS priorities to sensitive content like video calls, reprioritizing downloads and p2p applications, we could have video conferencing that works, 99.9% of the time.

[ryukafalz]

No. The FCC's 2015 open internet order had an exemption for "reasonable network management":

>A network management practice is a practice that has a primarily technical network management justification, but does not include other business practices. A network management practice is reasonable if it is primarily used for and tailored to achieving a legitimate network management purpose, taking into account the particular network architecture and technology of the broadband Internet access service.

Net neutrality has always been about leveling the playing field, not about preventing network operators from improving service for users. So, for example: QoS rules to make real-time communication work better in general? Great! QoS rules to prioritize Zoom over other real-time comms because the ISP has a deal with them? Nope.

[jack1243star]

Yet there's not one implementation that does work? I am not an expert on this, but I don't think current Internet-based solutions can compete with telephone/cellphone voice call on reliability, not without some drastic infrastructural change.

[criddell]

I thought a lot of phone calls were actually routed over the internet?

[mikequinlan]

"640K ought to be enough for anyone" said by someone, possibly.

If we have 10GB networks, applications will be found to make use of it. Possibly (hopefully) applications that we haven't yet even imagined.

[kortilla]

But the point is that applications haven’t even found a use for 100mbps to the home. I have 1gbps to mine and about the only time it’s been useful is when a friend wants to play a game I have on steam but not locally.

[mikequinlan]

4K TV (from Netflix) recommends 25Mbps for each stream. I know that this isn't 100Mbps but Netflix supports streaming to 4 devices at once (with the premium account). You also have to allow for other simultaneous usages (your kids playing games while you are streaming video, etc.).

This doesn't answer the question of what 10Gbps would be used for, but there are at least some applications today that use a significant portion of 100Mbps bandwidth. This is just speculation on my part, but history suggests that once 10Gbps is available inventive people will find a use for that bandwidth.

[MaxBarraclough]

The counterpoint to this is that once you've got enough bandwidth (and low latency) to stream an interactive session with negligible quality issues, you can move anything that requires more bandwidth offsite ('the cloud') and just stream the audio/video output.

As things stand though, it's as kortilla says: we don't even have any such bandwidth-intensive applications in the home. 4k video streaming is the most bandwidth-intensive problem we have. 10gbps may improve download rates for large archives, like modern games or perhaps major OS updates, but that doesn't strike me as a very compelling selling-point. It's a relatively rare occurrence, and reducing the download time by an hour (say) isn't worth reworking your Internet infrastructure. Even here, there's no gain if the Internet connection speed exceeds the write-speed of your storage hardware.

edit It's not quite the same, but I think 5G faces a similar problem. What's the point? 4G is more than enough to stream video. The biggest issue with 4G isn't the bandwidth, or the latency, but the coverage, and I don't think 5G is going to help there. (There are still parts of central London without reliable 4G coverage.)

[anongraddebt]

There could be some semantic issues here. It would seem that a non-trivial number of people have a connection where the actual speed differs a decent amount from the stated/expected. This has certainly been my case after living in L.A. and Chicago.

Perhaps the previous commenter perceives a 10gbps connection as the speed of an actual, say, 5gbps connection.

[MaxBarraclough]

I don't see your point here. For the purposes of our conversation, 5gbps and 10gbps are just the same; they're far in excess of 25mbps. Dishonest ISPs are another topic entirely.

[mikequinlan]

I don't want to belabor this, but you are saying "all applications that I can think of can be implemented using existing bandwidth". I am saying that clever people will invent new applications, ones that no one has thought of yet (or that no one thinks is practical), once the bandwidth is available.

[kortilla]

But the bandwidth is available and it’s not being utilized.

[owenwil]

Anyone who thinks this hasn't had to deal with a 100mbps connection before. It's not about a single user consuming all the capacity – though that's definitely a factor as well – but allowing simultaneous use of multiple demanding applications by more than one person.

[fastball]

3D video at 4K+ requires a good chunk of 100mbps.

[harpratap]

Streaming Games? Streaming VR and AR content? Lower latencies will allow more seamless video conferences. Right now our interface planes are still very 2D, which still has a steep learning curve compared to our 3D world.

[bawolff]

Sure, but latency and bandwidth are different things. Reducing latency by a few orders of magnitudes would open up tons of applications (not sure i agree that video conf is one of them, dont think latency of internet is the main issue there) but i don't think we are likely to see much beyond incremental improvements (light only travels so fast)

[fastball]

More bandwidth == less switching == lower latency.

Also light can travel from point A to the diametrically opposite point on the surface of earth in 66ms.

Currently it takes 227ms to ping New York from Tokyo[1]. As light travels, those two cities are only 36ms away from each other. Seems like there is plenty of room for improvement, even given the speed of light.

[1] https://wondernetwork.com/pings

[bawolff]

That's within 1 (base 10) order of magnitude of the theoretical max (which seems like it would be very hatd to get close to). Sure there is room for incremental improvements, but we're not going to be making order of magnitude improvements.

[fastball]

We already have the actual information traveling at the max speed, the delta is purely the cost of indirect routes and switching. It seems very reasonable to assume we could get much closer to the theoretical max.

Also this is obviously only a max speed if we never figure out superluminal communication, which... [fingers crossed]

[FartyMcFarter]

VR may require higher bandwidth too, in order to stream at 90-120 frames per second.

[owenwil]

I think we'll always find ways to use up the available bandwidth, even as it expands to new highs – but the real limiter has been how inconsistent high-speed deployments are, so it's difficult to reliably use them today.

I have a 1.5 Gbps synchronous connection here in Canada, but a few blocks away the max is 100mbps; we're not optimizing/building for fast connections because they're not widely deployed, not because we can't find ways to use them. Being able to play Stadia at ultra-low latency in 4K HDR is magic and shows a connection like this' value.

Making telco networks a public utility would help in this regard, especially if they have an actual goal of connecting everyone equally. In New Zealand, where I grew up, we broke up the telco and made fiber connectivity a public utility (chorus.co.nz) with a mandate to install fiber for free, anywhere in the country, which all ISPs can leverage. It not only increased competition, but it's helping create consistency, and will allow developers/content creators to dream up apps that can consume that type of bandwidth, because there's enough ultra-fast connections to actually make it worth bothering with.

[User23]

I was pretty happy with my 5mbps DSL in 98. Cable was significantly faster then too. To say nothing of the resnet. Calling that the dialup era isn’t exactly wrong, but early adopters were well past dialing into aol, compuserve, or prodigy by then.

[m463]

> latency

I think this is where we need to go.

so many experiences will become transparent with decent latency.

Not just usable, transparent.

It's like typing at the command line. If delays are less than 0.1s then a program is "interactive" and doesn't disrupt your thought process. Once you start noticing the delay you get poor results.

[birdyrooster]

Bandwidth doesn’t make up for the lack of good AR/VR equipment to make use of it.

[TomMarius]

In 2005, 10 Mbps was already the norm, and in 2010, we had 100 Mbps up and down.

[richardjdare]

This is quite a galling discussion :) I get ~2.5mbs. I live in a village about 30 miles from the second biggest city in the uk.

I've really noticed how ordinary websites have become sluggish over the past few years. To their credit, Netflix is ok as long as nobody else in the house is doing anything online.

I'm thinking I'll have to start some kind of local action group to get BT to take notice and install fibre.

[TomMarius]

Even wifi can be made way faster, why don't you start a community WISP? Small nonprofit WISPs basically formed the market in the Czech Republic, which is why I had these speeds.