Physical meetings will never be replaced

Image by 윤재 손 from Pixabay

Have you ever had that feeling in a Teams meeting where you interrupt each other by starting a new sentence “simultaneously”? And then you realize your interruption, so you go quiet, wait a little and start again because it seems as if the other person is giving you the que to go ahead, only to interrupt each other again. This almost never happens when you are talking to someone in the same room. I think the crucial difference here is latency. Over Teams, you introduce a handful of precious milliseconds, and they can make a world of difference. This is analogous to playing music in a band. To play in time, “tight”, the latency must be very low. This is also the reason why it is almost impossible to play music together online.

One of the purposes of a conductor is to make sure that the orchestra plays in time. An orchestra can be very large on stage, which means that there will be a significant distance between the musicians who are the furthest apart. When one guy plays a note, sound will travel across the stage at 343 m/s to the other side and reach the next guy’s ears, who will respond and play accordingly. Then the newly produced sound must travel all the way back to the first guy again. The larger the distance, the more time it takes to make a musical round-trip. It might be a bit subjective of course, but the physical limit for playing in time is around 8-9 meters. Any longer than that, and latency starts to become increasingly problematic. The speed of sound is 343 m/s which equates to ~3 ms/meter. A round trip at 8 meters distance thus becomes 30 ms. I am convinced that any practicing musician will be negatively affected by a 30 ms latency. A conductor is a visual timekeeper for the orchestra with no latency (except the speed of light), and consequently allows for comfortable playing at distances >9 meters.

In the recording studio, I personally get annoyed if the latency goes over 3 ms when playing electric guitar (i.e., 1 meter). I find 5 ms very annoying. This means that when playing a guitar amp simulation with headphones, it feels natural to me when the sound is delayed by about one meter. That is a pretty good approximation to where the amp and cabinet would be standing in the rehearsal room if I was playing with a band. Coincidence? I think not. I suspect that when playing the drums, latency becomes even more important due to the impulsive nature of the instrument. My guess is that the latency limit is the same as the distance between your ears and the drumheads. Any longer than that, and it would feel as if you had a neck that was 1 meter tall, and your ears placed high above the drum kit. Equally unnatural.

Let’s go back to the topic of video conferencing. Some googling tells me that a desired target is <150 ms for a good experience. This is way, way more than when playing music. If I personally can detect latency differences between 3 and 5 ms when playing the guitar, then no wonder why conversations can get awkward in Teams. We need to adapt our conversation style a bit to make it work, by expecting longer breaks between the speakers’ sentences, to safely start talking without interrupting. Take it slower. For a meeting where we take turns doing monologues, it won’t be any problem at all. I think the problems are much more common when talking 1 to 1. That is more like playing music. Humans are insanely good at detecting the smallest changes in body language, eye movements and muscle movements in the face. Remember, latency affects the eyes as well.

So, how close can we come to a perfect social interaction online? Let’s consider the speed of light 3x10^8 m/s and do a little though experiment when I am talking with my colleagues in Malmö. The distance between Umeå and Malmö is 1000 km = 1000 000 meters. The corresponding latency of a round trip at light speed is 3*2 = 6 ms. Okay, so that would correspond to having a conversation two meters apart (1 m apart should be more realistic). Albeit noticeable, that’s not half bad actually. But this is only the speed of light and the theoretical lower limit. We must include the total network latencies from routers and switches along the way and consider that fibers aren’t laid straight along the path. I don’t know exactly how much latency must be added in a real-world scenario between Umeå and Malmö, but I guess a total latency in the tenths. Either way, I think we can safely say that there is no, and there will never be, a substitute for physical meetings. Because it defies the laws of physics.