Quote from: fred21 on Today at 03:56:14 PM

What type of bug can there be?

There can be any kind of bug as JS interpreters and JIT compilers are complex pieces of software, same applies to any modern browser. And it wouldn't be the first time that there could be bugs in the PRNG. You seem stubborn to acknowledge that there could be bugs and you simply can't dismiss the existance of bugs.

Quote from: fred21 on Today at 03:56:14 PM

I can't see why some seed phrases are less secure than other.

You have been warned, but you're free to go ahead and ignore any advise.

Quote from: fred21 on Today at 03:56:14 PM

For example,I could write the 2048 words each one on piece of paper. Put all in a bag, then I pick 23 times a piece of paper by placing back the paper inside the bag each time. Between each picking, I shake the bag several minutes.
I do the same to get the 24th word by taking into account the checksum

Thank you for this example. It's great to show issues with trying to create true randomness, which is not easy, btw.

This might work well enough, but you can't be sure. Why is that? Well, let's assume your pieces of paper are mostly equal. If they aren't, the heavier might have a tendency to be faster at the bottom of the bag, the lighter might be statistically more often at the top of the pile in the bag. That scews the uniformity of distribution of the pieces. Can you quantify it? I guess not really. It should concern you though, not to know how much of influence this makes.

Equal flat pieces of paper might tend to stick together, maybe only two pieces of paper stick together but then maybe you end up with multiple pairs of such stickies. If they only stick together for some time during shaking, it again hinders equal distribution of the pieces during shaking.

How do you want to make sure that the way you shake gives you an equal re-distribution of the pieces?

Next variable is how and from where you grab your piece when you draw one after shaking. If you distort the piece then you change it's behavior compared to the rest of the pieces. Then not all pieces are the same anymore.

I'm not sure if you get it, but understand that what sounds to be random is actually not easy to be sure it's truely random. And for the seed of your wallet, the random entropy of 128 or 256 bits, you definitely want the entropy to be as random as possible.

Quote from: fred21 on Today at 03:56:14 PM

By the way is there a way to measure the randomness with which a seed phrase was generated ?

I'd say, not really, as the sample space of at most 256 random bits isn't enough to get good answers by the randomness tests that exist.

Usually you test your RNG source thoroughly with large sets of samples with randomness tests (there are various) and to call your RNG a good one it should pass as many test as possible without obvious flaws. Only then you know that you can rely on your RNG to produce good random entropy as your entropy seed for your wallet.