Hours collected vs hours logged.... data processing time

Hi All,
Just a discussion topic with no real motive other than just speculation, but…

The other day, after a couple months of no new devices, I got an e-mail at 1am saying that a new unknown motor had been found. I checked the Sense monitor “status” screen to see if the unknown motor under currently learning was gone, but found that it was still there and had not changed (Which it hasn’t in quite some time). But now under the “up next” box, there was a microwave.
This made me think about just how much data they are ingesting and what their computational power must be to handle all the data. Thinking back to the Seti@Home days where they would talk about how far behind they were on processing data.

Today on Facebook, a user posted a screenshot of some logs from his Sense he received from Sense support. What caught my eye was that it had two lines. Hours: and Hours Processed. In his case, there was only a difference of 14 hours, but I wonder if with all the additional Sense users, if they are getting behind on processing data. @Howard recently had to start over with all new data and commented that he received some initial results much faster than his initial discovery. Now this could be due to a whole combination of reasons, but I wonder as well if new users are getting priority on server time to try to get them some results quickly, and then it peters off and goes into the general cue.

I sort of assume that the basic premise is that one a device and its associated power model is identified to a house, they can send that model to the local Sense and it has enough processing power to detect those on its own without having to go to the cloud. Otherwise, it implies that the “cloud” processing is always required and as users grow, computation hours will increase which will cost more, and ultimately put Sense out of business as without a subscription service, just the sales of the Sense unit would not sustain the processing costs unless the cost of processing each “data unit” goes down at the same rate as they sell and onboard new users.

Not trying to bash the company or imply positive or negative methods they use, just some food for thought / comment. Even though my Sense hasn’t identified many devices in my house, even just as a whole home power monitor, it has helped me make changes.

You bring up some interesting points about the business model.

Many (most) of the devices in my house haven’t been found so I’ve been trying to figure out how to use the information that Sense gives me, such as the immediate usage of an appliance coming on (not discovered) and the over-all usage of the day.