The compute times for the proof-of-work varies considerably within a random sample of input data. With some input data, the POW can take 1/100th to even 1/1000th as little time as it takes normally (and vice-versa).

Given that the POW time varies considerably, is it possible to create transactions that take advantage of this by explicitly constructing them to complete proof-of-work in less time?

1 Answer 1


It should be impossible.

To successfully accomplish the POW, you must find a nonce (i.e. a random number) to append to the transaction data so that the hash of the whole data (transaction data + nonce) ends with at least N zero trits (the hash in trytes will end with '9' tryte) (N=14 on main net).

The hash function is designed in such a way that it is a one-way-function: you cannot deduce the input from the output. Or more generally, you cannot deduce input-requirements based on output-requirements (output-requirement = ending with 14 '9' tryte).

So the only way to accomplish the POW is to try different nonce until you find a suitable one. If you are (very very) lucky, your first attempt can be successful and you do the POW in no time. If you are unlucky, it can take ages before finding a valid nonce. But it the end the average time to find a suitable nonce should remains more or less the same.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.