(home)

My experience using Google Meet, Zoom, and Discord video calls has been mediocre. Obviously they're going to be worse than in person, which features ~5ms latency, high quality, and an absolutely unreal amount of nines.

But there's something more fair to compare them to, POTS (Plain Old Telephone Service).

In the most old-school instantiation of POTS, your voice is turned into an electrical signal and then rocks down the cord at 200,000,000m/s, straight to the other party's phone. That's pretty cool. I've read anecdotal reports that they created a much better audio experience than video calls. When I search my memory that seems right. But I'd like some data.

Dimensions to this Question

• Aspects of call quality
  • Latency
  • Jitter
  • Sound fidelity
  • Duplex status (see this thread)
  • Misc
    • Such as anti "call-yell" tricks like sidetone
• Situations
  • Long distance vs. local calls
  • Change over time
    • At some point POTS providers stopped using direct connections and added digital links internally for long distance calls. How did this affect things? Did local calls over a landline continue to use direct connections?
  • Multiple different video call services: Google Meet, Zoom, Discord, etc.
    • Also interested in apps that are pure audio like Mumble
  • Multiple things involved in video call quality. Eg if latency or reliability is bad, which of these is usually the contributor:
    • The internet providers of the participants
    • The video call servers
    • The users' computers and video call clients

What I've Found so Far

• Can You See Me Now? A Measurement Study of Zoom, Webex, and Meet
  • arxiv.org/pdf/2109.13113.pdf
  • 2021
  • This part is for the video not audio, but still interesting:

    • In the US, typical streaming lag experienced by users is 20–50 ms for Zoom, 10–70 ms for Webex, and 40–70 ms for Meet. This lag largely reflects the geographic separation of users (e.g., US-east vs. US-west).

• HN question by me
• HN thread on the article A Theory of Zoom Fatigue
• 📝 contact me if you have more suggestions!

Possible Practical Applications

Obviously the country isn't going to switch back to landlines, but I can still think of some uses for answering the above questions.

• If poor video call quality is due to varying sources we should do a better job of surfacing which is the problem for particular calls.
  • Right now users often don't know on which end of a two-person call problems are coming from, much less whether the issue is with the video call service or their Internet provider.
    • If the latter they need to know whether it's a latency or bandwidth issue.
• Google has Project Starline (HN discussion). Should more businesses consider latency optimized video call clients?