Taking a slice of a VRF RandomWord - still random?

[tmcdonough]

If you take a slice of the randomWord returned by VRF, is that also random? E.g. say vrf returns:

105409183525425523237923285454331214386340807945685310246717412709691342439136

If i grabbed the first 2 digits and the next 2 digits (e.g., 10, 54), would those two numbers each be random, as well?

[othaime-en]

Hmm...Good question. Well my thinking is...if the entire number is random, wouldn't the subset be random too since it would be hard to predict what these two numbers are if we don't know the entire random number?
What are your thoughts? @PatrickAlphaC @alymurtazamemon @krakxn

[krakxn]

Why not just use math.random via the ceiling value and *12 for a natural number (b/w 1 and 12).

[tmcdonough]

i thought the reason we use VRF is to ensure a truly random number, rather than generating via the blockchain which is deterministic? I need the "dice roll" to be a random number

[krakxn]

Yes, a VRF essentially serves as a function to generate such a random number. But in your case it seems to be a "dice roll", which probabilistically is what I mentioned above, with a 1/12 probability to predict which side will land. Using a random function--in this case--defeats this purpose.

Of course, for something like a lottery (and similar applications), you need it to be truly random--hence the usage of VRF.

[tmcdonough]

Say I have a function "roll" that costs 0.1 eth to spin. Calling this function uses math.random to generate a random number between 0-11. 1.0 ETH is paid out when the "roll" returns the number 6.

Since this will return a number that is not truly random, couldn't a clever user wait to call the "roll" function until it was expected to generate a 6, ensuring they'll always win the 1 eth?

[krakxn]

"not truly random" It is still random, it is just that the range is 1 to 12, hence the probability of guessing increases significantly relative to a random function like VRF. Of course, using the latter would be the ideal way for randomness, but that certainly is not representative of a 12-sided dice.

They certainly could; you need to consider game theoretic viewpoints of all the participants such that it is not in their economic benefit to do so whilst considering Sybil attacks, which would be prevalent in this case.

Seemingly, this is a bad idea as far as scalability is considered; of course, for fun w/ friends it is all good.

[alymurtazamemon]

@tmcdonough Like the hash functions that change the complete hash key by just changing a litter thing in it. Ex. giving space or adding a .. These random number generators use the same hash functions to do so.

But randomness does not mean that the number/value will not come again, obviously if you choose from [1,2,3] array, and run it multiple times definitely the chances of returning the same number will be high.