Adjusting the Default Price of Boost

[owenkellogg]

The problem is to estimate a good default price of Boost POW in sats / difficulty.

The idea is to look at how long users have to wait to get their boosts mined. If they have to wait too long, then the price is too low. If it is too short, then the price is too high.

We target a value in seconds per satoshi, which is the amount of time users have to wait per satoshi spent. We look at all boosts generated at the default price over a given period, say from two days ago to one day ago. Of those boosts that were redeemed, we sum the time taken to mine each and divide by the total satoshis spent on them.

Let P be our default price and let T be the amount of time we expect we expect a user to wait per satoshi spent. Let X be the measured value of wait time per satoshi.

The new price of Boost will be P * (1 + c * (X - T)), where c is some small value.

The default amount that users pay, which currently is 5 cents, should be a constant value that we set, so what should actually change is the amount of difficulty that is bought for that value.

[DanielKrawisz]

Where I say c is some small value, that means less than one. If the value was too small then the price would adjust too slowly. If it was greater than one or too close to one it might lead to instability, so something around .5 or .75 ought to be good.

[owenkellogg]

This algorithm still needs to account for jobs created at the default price that were not mined in the previous 24 hour time period. If only 1/3 of the jobs were mined then we should multiple the expected time to completion 3.

[create-issue-branch]

Branch issue-38-Adjusting_the_Default_Price_of_Boost created!