Hacker News new | past | comments | ask | show | jobs | submit login

That's impossible. Look at the Nest thermostat - trying to learn when people will be home. It doesn't work beccause most people live messy, noisy lives.

Even I don't know when I'm going to bed until just before I do so. I might go upstairs to brush my teeth, but then go back down to play more video games.

I might go into the bedroom to pick up a book - doesn't mean I'm going to sleep.

The idea of predictive analysis of human behaviour is almost nonsense. Even if it weren't, a quick shout to a robot, or a tap on a button, is quicker and cheaper.




The fact Amazon cannot be more stupid with its' recommended goods than it is doesn't set the fact you just cannot predict someone would want to buy laser printer cartridge in a month or so after he bought the printer.

You will be surprised how repetitive your patterns (like pulse, for example) are and how well they predict will you decide to play the game or to go to the bed. Maybe you just don't analyze these patterns thoroughly enough, so you have an illusion of unpredictability of your decisions.

And by the way, smart things aren't there to only predict. If your blanket, and all other devices - including your todo list app - would vote for the bed, especially if they formulate it with carefully crafted suggestive words in sweet voice, you better agree.


Not OP but ... If you turn off the downstairs lights && go upstairs && it's after $time ... ?

That would work mostly for me, perhaps including the TV and computer being off. That's going to cover 99% accuracy, probably as much as I'd get with a button.


The smart blanket would predict it much earlier, to have the time to warm it up.


I am not sure I understand the benefit of predicting it that early. My electric blanket needs only a few seconds to heat up


I haven't set the temperature on my nest thermostats in over a year.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: