Proctorio - There's a snake in the house

Midterms are upon us, and with them comes all the possible solutions to proctoring in the time of Covid. In classic Selkirk style, the solution wasn’t presented until two weeks before midterms and even then we were not able to actually use it for a while.

Last week I was the first to test it on actual students. It was an epic disaster. I was sucked into a complete mess of a week and by the time I did get it working properly I felt truly defeated. My extensive interactions with the support team were underwhelming to say the least.

But getting it to work was actually the tip of the problematic iceberg.

Proctorio is a Chrome plug in. In order to work it must be enabled on all sites. It’s permissions include:

• Read and change all your data on the websites you visit

• Display notifications

• Modify data you copy and paste

• Capture content of your screen

• Manage your downloads

• Identify and eject storage devices

• Manage your apps, extensions, and themes

• Change your privacy-related settings

Disturbingly extensive, don’t you think?

In addition to requiring this level of access, it also is required to be allowed during incognito browsing.

As a private company, I have no idea what Proctorio is doing with all this data, but given that it is a tech start up, I’m guessing something.

The next issue is the level of data it takes from students during an exam - a copy of their ID, a panoramic photo of their room (multiple times throughout the exam), the exact longitude and latitude of the writer, their IP address, and a plethora of biometric data. One exam would provide enough data to keep a home burglar in jobs for the next year.

However, my gravest concern is the suspicion score. Based on AI and biomentric data such as “abnormal head movement”, each student is given a suspicion score between 1 & 100%. I’ve had very few students clock less than 50%, and I can’t actually drill down into how this is computed. For me, this makes this data useless, but my concern is that if someone should try and rely on this data, what kind of bias is baked in? Are the movements of neuro-divergent students on the spectrum “abnormal”? Are the head wiggles of Indian students “abnormal”? Given that the staff of this company appear to be primarily white men, I have to believe that these scores are not fair to anyone who is not, well a white man.

In short, this software scares the bleep out of me.