Hacker News Reader

I asked this last time about the one in London, and was told that one of the checks is that the image has changed since the last run. Otherwise the data isn't used.

Prevents not only technical issues but attacks like someone blocking the camera or putting a static photo in front of the camera.

Interesting - sounds like it would have some negative effect then. Thanks for sharing.

Now I wonder about some periodic offsets. E.g. if the lights are off at night, or if the skies are overcast in winter, does it skew the results in some significant way. I seriously doubt that though.

Not really. It's just one source of randomness among many. The entire point of having multiple sources is that they are redundant, you don't need them all.

Surely not. If you're seeding a PRNG from multiple sources of entropy, you generally concatenate them. Or if you were limited in bytes you'd XOR them.

This is why, in an app, you might seed with timestamp and process ID and /dev/urandom, in case any of them happen to be non-unique or unsupported.

a random number (existing entropy) XOR with a static number (the crashed wall) is still a random number, me think

Probably not, unless it is their only source of entropy.