20ms


The Cybercultural Singularity

Sometimes I feel old for the computer and Internet industry. "Old" is thirty five. I first visited the Internet in 1993, and I remember bulletin board systems. I talk to people on tech startup boards who were born only a few years prior. Ouch.

It was a different cyberspace then. There were toll booths on the roads in the form of long distance calling fees. These made BBSes largely local affairs of same-area-code users. A long distance caller was like a bedouin coming in from the desert. These strangers brought exotic trinkets and strange ideas from far away area codes: hacker 'zines from other cities, the latest "warez."

There's a site called textfiles.com that hosts a decent collection of artifacts from this prehistoric culture.

Today the Internet is largely a dimensionless point in space, a singularity of a different sort where all information and culture is condensed into formless unity.

It's a triumph in engineering, and it's brought untold riches in the domain of information. It is now possible to query almost the whole of human knowledge in seconds and to retrieve it whether it's at your local university or on another continent.

Yet it's also brought a few downsides. With the removal of the distance barrier has come the transformation of the online world into a monoculture dominated by snark and "memes." It's also removed an important aspect of that old localism, namely that online relationships could easily translate into offline ones. Since BBSes were local affairs, BBSes had meetups where you could meet the people you were talking to and physically interact with them. Many great friendships were born in this way.

Sure it's possible to have meetups via sites like Reddit, but somehow it isn't quite the same. The shape of the medium affects the culture that forms within it. Reddit's culture is Reddit's culture, not the culture of any actual place.

It Will Happen in the Matrix When They Sync Something

I've been wondering for some time now: the web feels played out, apps are rapidly becoming same, so what is the next virtual medium?

It's starting to look like it might be an old idea, but one that advancing technology is finally managing to lift from the sci-fi novels: virtual reality.

Now that beta headsets are on the market, people are starting to think about what we might build in this new media space.

I had a bit of an epiphany when I read a few of the articles on a new site dedicated to VR design principles. The site is called Twenty Milliseconds.

20ms, it turns out, seems a bit of magic number. If the round-trip latency between an action (say, moving your arm) and you seeing that action is longer than ~20ms, the result is a "loss of immersion," possibly even nausea. The universe is pretty low latency. Our brains are simply not configured to deal with the weirdness of sluggish causality.

The notion that latency matters -- possibly a lot -- in VR design is interesting because the Internet is a machine.

Remember what I said about the online world having collapsed down to a singular point in space? It's not true. Try pinging a system in your city, and then another on another continent. The further a system is from you, the longer it takes for a message to get there. We don't usually notice it very much because a difference of a few hundred milliseconds in the load time of a web page doesn't really matter subjectively.

A web page is also a mostly static entity. Once it's loaded, it's here and we can scroll around in it. Apps are the same. Once you download the app, most of the data is there on your device. The interactions you make via the app with other entities out on the network may take anywhere from a few milliseconds to a few thousand, but since those actions are mediated through a non-immersive interface your brain doesn't care.

Latencies on the Internet are physical. I'm sure it would be possible to squeeze them down by making routers and switches faster, but ultimately you're going to hit things like the speed of light and routing performance limitations deriving from things like queueing theory.

How much this matters will depend on how rich we want our virtual worlds to be. If they're mostly static universes, it won't matter much. It'll be possible to pre-cache everything on your local system or on very nearby servers.

But if we want our interactions to be richer... and we almost certainly will...

Picture two people sitting across from one another at a virtual table. Before them is a virtual donut. They are going to fight over that donut. One grabs for it, then the other. What happens?

I'm in Orange County and I have a server in Los Angeles. My round trip ping time to that machine is about 4ms. If my donut-wrangling partner were in, say, Encino, our interaction would take place fully within the 20ms time horizon. The physical behavior of our arms as they collided with the donut and the motion of the artifact itself as we struggled over it would seem "natural."

But if that person were in New York, or Amsterdam... I picture a donut herking and jerking about, disappearing and reappearing as some database synchronization algorithm struggles to maintain consistency between two discordant bubbles of consensual hallucination.

Now picture the streets of a metaverse, something as rich as what Neal Stephenson described in his VR novel Snow Crash. In Snow Crash Stephenson nods to the physical limits of networks, describing how virtual physics would decay when too many avatars jammed into the same field of view. He described it in a way that made the metaverse just bizarre and dreamlike, but what if the real effect of coherency breakdown is less interesting?

What if inconsistency and sync artifacts are just flat out hokey and ugly? What if any perceptable time delay just destroys the sense of immersion, or even makes you want to rip off your goggles and run for the bathroom?

In the Matrix films, they told Neo that deja vu was the sense you got when you were in the Matrix and they changed something.

If your VR peers are not nearby, you'll be experiencing deja vu pretty much constantly.

BBSing 2.0

Us coders are pretty clever. Maybe we'll figure out mixtures of virtual world limits, interface metaphors, and sophisticated algorithms to render the latency barrier less relevant. But I wonder: what sacrifices will have to be made in the form of richness, realism, and world mutability?

Such sacrifices are normally made in MMORPGs and we tolerate them there. But those are still played on a screen. They're not truly immersive. If you look at how people use MMOs, they treat them as sort of like a mash-up between a video game and a chat room.

Will immersion be the same? Or will we want... you know... real immersion?

If we're not willing to settle for still life, the 20ms time horizon might enforce a new physical localism.

I think that would be very interesting.

Picture a metaverse peopled by avatars and artifacts from your physical region. Now picture a long distance traveler, a client connecting from distant London, Shanghai, Sydney, the ISS. The visitor's avatar is translucent, a ghost from afar, able to speak but not physically interact.

He or she is obviously not from around here, intrinsically of interest yet also intrinsically other. Perhaps translucency equals latency. Perhaps when they pick up an object it also becomes translucent, temporarily under the influence of a spirit from a different geophysical reality.

The opaque are but a short journey away. You know each other in real life. The virtual blends with the physical.

It could be a very interesting medium indeed.

About the Author

I'm a software engineer and tech entrepreneur who is presently nursing along a bootstrapped project called ZeroTier. I think a lot about all kinds of things. Lately I've been thinking a lot about the network as a medium and how the shape of that medium affects its culture. This is my personal blog, so what I write here represents my opinion alone and blah blah blah.

Comments

Show Comments