Quite a lot of the specific details did surprise me though, to the point of wondering if there's a bug somewhere in their code determining percentages. "Audio context": "sampleRate : 44100, state : suspended" 0.05%? Gyroscope is a green check (I don't even 100% know what that means) and that's 0.13%? "Battery": "charging : true, chargingTime : 0, level : 1", 0.08%? (And is that even useful for fingerprinting over any significant period of time anyhow?)
So seems very skewed!
There are a few studies that used sites like that and got lots of traffic (by advertising on sites like HN) from non-normal users.
There was a landmark study done in France that measured fingerprints through a government website. The general population was much less unique.
Edit: found the AmIUnique paper: https://scholar.google.com/scholar_lookup?journal=Proceeding...
If I had to guess, based on what I know about you (what I mentioned in the previous paragraph), I would guess that you built your own computer. This almost guarantees a unique fingerprint.
With the deprecation of 3rd party cookies, we will definitely see an increase in fingerprint-based tracking. That being said, it’s going to look different than unique cookies (and it doesn’t explain OP’s situation.
(Or it might actually be that simple. Your fingerprint might be unique between IP re-addressing, and that’s all it is.)