May 11 Day 64: Scientific Method

Ok, I admit it, participant placement in Zoom meetings is becoming a bit of a personal obsession. What started as a puzzle (May 2 Day 55: Start Small) with a cute solution was tested in a slightly larger setting (May 5 Day 58: First Approximation) and found wanting.

From observations made while teaching to the many “white-named-black-boxes,” it appears that people streaming a camera feed precede people not using the camera at all. And then I remember during the meeting with colleagues on May 5, Ander stepped away to take an important phone call. I can’t recall if his camera feed was stopped temporarily but that could be a plausible explanation for his change of placement especially if not everyone took their screenshots at the same time. So my prediction that positions are determined simply by the order in which people join the meeting does not take into account dynamic interactions as we turn cameras off and on over time.

A new hypothesis. I predict that Zoomers are placed in the gallery view participation grid from left to right, top to bottom, in the order that they join the meeting subject to the following restrictions:

• Participants see themselves in the second position.
• Participants with active cameras occur before participants with inactive cameras.
• An active participant that mutes their camera moves to the inactive portion of the list (and I haven’t a guess as to whether it should be the front or the back or somewhere in between.)
• An inactive participant that turns their camera on moves to the end of the active portion of the list.

Empirical Testing. My mission at the next large faculty zoom meeting was to pay particular attention to placement in the gallery view as I (and others) turned on and off cameras. I signed into the meeting on two devices: my computer and my iPad. The meeting had nearly 100 participants, so I hoped no one would detect or be bothered by my experimentation. I muted the camera on my computer and you can clearly see the same avatar in the second position of every image below. I took screen shots on my computer while I toggled the iPad camera on and off. Some results are presented in the gallery below.

• 

• 

• 

Results. Based on the first image, I should have been the tenth person with an active camera when I joined the meeting. In the second image, the black box “Jenny Quinn’s iPad” is the first position after the active feed. And in the third image, with the camera back on, I moved to the end of the active line. There is one static box (Will Burghart’s Don’t Panic image) ahead of me in the line of active camera feeds in image 2. You can see from the yellow box around Will’s avatar that he was speaking in the first screenshot so perhaps that gives him precedence over people with newly turned on cameras. I observed Sarah Hampson turn on her camera and pop in front of my black box between image 2 and image 3. I continued testing and monitoring throughout the meeting and I believe that the hypothesis above is close but it doesn’t take into account how the use of microphones impacts placement (or for how long).

It’s time to reformulate and iterate. The next tests will need to be completed under more controlled conditions. Questions continue to arise like how does the algorithm break ties if two or more people arrive at the same time? When you are automatically sent to breakout rooms, what order are participants placed in the room as everyone is arriving at the same time? And what is the precedence and interplay of using sound and images?

Many of my friends and colleagues are finishing up their semesters; they get a respite from daily Zooming for a little while. I still have four long weeks of teaching to keep contemplating how Zoom decides placement in my classroom and elsewhere in my pandemic life.