General statistics
List of Youtube channels
Youtube commenter search
Distinguished comments
About
Kristopher Driver
Computerphile
comments
Comments by "Kristopher Driver" (@paxdriver) on "Computerphile" channel.
Previous
1
Next
...
All
a live demo would've been peeeerfect in this video ;p thanks guys, love your work
16
@HMan2828 we have atomic clocks syncing time all over the world by satellites and atomic decay or resonance, but we don't have a proper random number generator network for the world's digital security producing truly random number by a camera staring at a lump of plutonium or something? How is a world random generator project not a thing yet? It's at least as important as going to mars lol randint api would be super helpful,, wouldn't it?
4
If I ever get a watch that tells me to calm down when I'm laughing too hard or kanoodling you'd have to give me a free car to wear it for that level of intrusion. Hospitals might buy it for patients I guess?
4
You should totally do an episode covering the nitty gritty of the proof. That would be awesome!
3
Brilliant!
3
Why not sum sliding window drifts of different sizes, then weight the functions by window size like ex moving averages used in market analysis?
2
DNA is innately recursive! ;p biochem simulations
2
@travelthetropics6190 they do, there should be a unified random number generator with oversight and transparency by independent groups though. A central service like GPS but which provides random numbers for global security and personal use rather than rely on company implementations of random number algorithms. There should be a central source for proper random numbers imho. The cost benefit is massive. Small developers need security too and software users all benefit from unpolluted random number generation at all scales. It would be a cheap but useful global co-op initiate
2
Isn't this easily solved by making the rule "robot make sure I have access to the button. Robot can't get any reward unless I am able to press the button at any time." what am I missing?
1
1965 they weren't measuring transistors in nanometers, they can't scale down infinitely, hence the need for multiple cores, hyperthreading and bus speeds. My prediction would be that they'll need supercomputers to efficiently hard code these multicored super comps and we'll go back to algorithms making things faster rather than processing. The brain works because of its ability to rearrange itself on the fly
1
@TheSam1902 it wouldn't be the same result but they'd both be approximating the same function, except my method could be processed with streaming data and the other would need to be fed into tensors I think.
1
Isn't there a way to implement an onion routing pseudo method, maybe using PGP? Wouldn't that be easier to accelerate in 5nm chiplets too?
1
Social security number / government managed blockchain would work. Voting is no more important than finance
1
Was this related to Meltdown?
1
@AndersJackson they do demos all the time on this channel. C, C++, Python, I've seen several low level code demos on this channel, a short 7 or 8 terminal commands wouldn't be out of the ordinary at all.
1
@HMan2828 those systems are pseudo random, and mostly work but depend entirely on the implementation. For eg, most things in nature oscillate, so generating entropy in a circuit is actually very difficult because they're incredibly precise. They may cycle in modulo once in a hundred million but computer clusters can be scaled to process values in the millions of billions of millions . Statistically all pseudo randoms are inherently flimsy because fast cuda cores can be hand coded to process data in parallel with 4,000 tensor cores in a single retail gpu. Cross with that, running thousands per second (gpu clocks) means finding a resonant modulo of pseudo values within a few seconds if the pattern repeats once in a thousand billion iterations with some predictability. That's a characture of reality, obviously, but not by a lot. 1s of processing time is all that would take so clusters spending years watching its outputs eventually any pseudo random will be unravelled. .
1
@MicraHakkinen lol true, it would take planning. I'm confident a VPN direct from the facility would suffice. You don't need the perfect randoms to access the service, it would be for mass generating randoms like server to server. Also numbers can be buffered ahead of time and sent out in batches, there's no reason it has to be on demand if the user logging in to the service has a sandboxes 2mb of stored values. User could even use a pseudo random just to picks which real random to use. There's a million ways to sort it out or set it up as a service. Being scared of hackers makes no sense, you'd have to already be compromised for that to be an issue.
1
yeeeeees! great stuff
1
King Arthur, Archimedes was counsel or something - alongside Merlin, no?
1
Previous
1
Next
...
All