[ / / / / / / / / / / / / / ] [ dir / ashleyj / bl / girltalk / imouto / lit / rsty9793 / sg / tijuana ][Options][ watchlist ]

/tech/ - Technology

You can now write text to your AI-generated image at https://aiproto.com It is currently free to use for Proto members.
Name
Email
Subject
Comment *
File
Select/drop/paste files here
* = required field[▶ Show post options & limits]
Confused? See the FAQ.
Expand all images

Merry Christmas!

File (hide): df039d075fa1e58⋯.jpg (128.86 KB, 800x533, 800:533, 1423424893020.jpg) (h) (u)

[–]

 No.842219>>842231 >>842270 >>842274 [Watch Thread][Show All Posts]

So, i want to build my own RNG only with ICs, how hard this can be?

Can we trust that there is no backdoor on the integrated circuits?

Is that even possible?

 No.842231>>842318

>>842219 (OP)

>he can't cut open a IC


 No.842235

>implying his integrated circuit doesn't have a backdoor


 No.842236

Use a libre hardware design and base your work off that.


 No.842240

I'm pretty sure we already had this thread.


 No.842270>>842283

>>842219 (OP)

just xor a bunch of shit together. even if one of the inputs is backdoored it wont matter

>Is that even possible?

yes


 No.842274>>842280

File (hide): 5070cf4aa40bda1⋯.gif (5.63 KB, 480x240, 2:1, random.gif) (h) (u)


 No.842275>>842284 >>842884

Depends on how random you want it to be. Without using radioactive decay, you can just cobble together a bunch of sensors and have their outputs come together to form one number.


 No.842280>>843014

>>842274

>using backdoored transistors


 No.842283>>842289

>>842270

>just xor a bunch of shit together

https://blog.cr.yp.to/20140205-entropy.html


 No.842284>>842294

>>842275

>Without using radioactive decay

There are other quantum phenomena that can be used. Radioactive decay was just the first used, and they didn't even know how truly random it was at the time.


 No.842288>>842290

Also it's algorithmically possible to make certain kinds of unfair sources fair, like say, an unfair coin. Flip the coin and record the result. Then flip it again. If you get the same result, discard your recorded result. If it is different, retain the recorded result. Even if the coin is unevenly weighted, your string of heads and tails will be random and approaching 50/50.


 No.842289>>842290

>>842283

>But what if z comes from a malicious source that can snoop on x and y?

seems to be not applicable


 No.842290>>842304 >>842313

>>842288

Alternating heads/tails is 50/50 but not random. Are you retarded?

>>842289

How?


 No.842294

>>842284

I was more just referring to more complex systems.


 No.842304>>842305 >>842595

File (hide): 762e04d4814edb1⋯.png (326.77 KB, 1584x1786, 792:893, 7428ddea38254a86a8a18b9200….png) (h) (u)

>>842290

Where are you getting alternating heads and tails from? Let me try explaining it so even your feeble mind can understand.

Flip a coin twice. If it's the same result both times, ignore it. If they're different, write down the first result. Do this until you have as many coin flips as you need. This gives you fair random coin flips. They do not alternate. Try it. Put a couple layers if duct tape on one side of a penny to make it unfair. Then use the algorithm I described. It was developed by Von Neumann. Even if the coin in skewed so that the probability of getting heads is .99, using this algorithm makes the final probability of getting heads .5

It seems you are the retarded one.


 No.842305>>842308 >>842309

>>842304

Holy shit are you retarded? You literally said: "Also it's algorithmically possible to make certain kinds of unfair sources fair"

You propose to do this by discarding values if they are the same as the previous value.

If you have a source of "randomness" that produces alternatingly two values, your algorithm wouldn't discard any values.

>posting le epic furry reaction image

Holy shit go back to your containment thread.


 No.842308>>842310

>>842305

Unfair != nonrandom. Get your head out of your ass and go take a statistics class. Unfair means the probabilties of each outcome are not equal. Like for example the unfair coin and Von Neumann's simple algorithm to restore fairness.


 No.842309

>>842305

Tree Algorithms for Unbiased Coin Tossing with a Biased Coin

https://web.eecs.umich.edu/~qstout/abs/AnnProb84.html


 No.842310>>842311 >>842335

>>842308

But for cryptography you don't need a "fair" RNG. You need a random RNG.


 No.842311>>842312

>>842310

Not true. Also, changing the goalposts.


 No.842312>>842332

>>842311

>Not true

It is. For cryptography you need a unpredictable source of randomness. Not a fair one.

>Also, changing the goalposts.

OP was talking about backdoors. It makes no sense to backdoor an RNG if it is not used for cryptography.


 No.842313>>842314

>>842290

>How?

if you have one super leet quantum random decay IC which is just backdoored, you could still have 10 other sources of entropy and xor them all together. the backdoored IC doesn't have a way to read from these other ICs


 No.842314>>842342

>>842313

>the backdoored IC doesn't have a way to read from these other ICs

that is where you are wrong, kiddo.


 No.842315

>Using backdoored physics


 No.842318

>>842231

>implying the cutter isn't part of a botnet


 No.842332>>842334

>>842312

Holy fuck man. Unfairness literally is a type of predictableness. The fact that it can be statistically predicted is what makes it unfair. It's why using loaded dice will get you arrested for cheating at a casino.

Random numbers have three qualities that make them unpredictable (i.e. truly random.)

High Entropy: There are no patterns.

Fairness: Each outcome has an equal probability.

Non-replayability: You can't use past numbers to determine future numbers.


 No.842334>>842531

>>842332

Retarded nigger. Stfu if you have no clue.

An RNG that ouputs alternetigly 0 and 1 is fair but completely predictable.

>Random numbers have three qualities that make them unpredictable (i.e. truly random.)

Holy shit you are LARPing hard.


 No.842335>>842338

>>842310

random implies fair.

because otherwise you can always test the value which has higher probability and have more than 50% chance of guessing values on average, which makes it partially broken.


 No.842338>>842343

>>842335

>random implies fair.

The only correct thing you have said so far. I'm glad we agree.


 No.842342

>>842314

just don't have side channels?


 No.842343

>>842338

It was my first post in this thread dude


 No.842361>>842363

how hard would it be to get whatever RNG you come up with to interface with linux over USB or something? seems like your going to need a SBC for this unless you really know what your doing with EE shit and writing linux drivers and code for the circuit to actually talk to linux


 No.842363

>>842361

Just use the serial part / use that one ftdi chip which is serial to USB.


 No.842531

>>842334

An RNG that outputs zero and one alternately is not a RNG. Not even close. It has no entropy and is completely replayable. Like I said, fairness is a type of predictability, it isn't the only one.


 No.842595>>842658

>>842304

That's meaningless as in reality we're not dealing with bias, we're dealing with "random" sources that produce known sequences. The output of an encrypted stream of nothing but 'tails' would look like random, unbiased coin flips and a deterministic algorithm like that applied to it wouldn't change a fucking thing.


 No.842658

>>842595

> The output of an encrypted stream of nothing but 'tails' would look like random, unbiased coin flips and a deterministic algorithm like that applied to it wouldn't change a fucking thing.

The algorithm is for correcting bias when generating random numbers, not for manipulating already encrypted data.


 No.842884

>>842275

The inputs will end up converging to certain values over time.


 No.842910

just use RDRAND like the linux kernel


 No.843014

>>842280

>not librebooting your transistors when they arrive




[Return][Go to top][Catalog][Screencap][Nerve Center][Cancer][Update] ( Scroll to new posts) ( Auto) 5
39 replies | 2 images | Page ?
[Post a Reply]
[ / / / / / / / / / / / / / ] [ dir / ashleyj / bl / girltalk / imouto / lit / rsty9793 / sg / tijuana ][ watchlist ]