This is perhaps a little “meta-data analysis” but a recent post has me thinking
Would Sense benefit from deliberately failing some smartplugs and other devices to get some data? While this wouldn’t be ideal ML data it could set the bounds for searching existing and future data.
Is Sense amassing a “failure” library? Good vs bad device waveforms?
It would seem (much) better to have safe lab-based failure-data before the data is generated in the wild.
I think it would be hard to get “typical” failure waveforms out of smartplugs. But I would certainly want to fail some of the less expensive appliances in the lab… Pretty expensive to fail a few $4K central air compressors.
Agreed.
I was imagining the §prime alert suspects such as Li-ion charge failure and so on. The dangerous stuff first. That includes smartplugs.
This is distinct from fridge and HVAC fails which through their installed ubiquity would seem to be amassing usable failure data. And generally (I hope) their failures just cost money vs. cause fires.