RE: Policy Parameters for a Community Bot
It all looks really good, but I have a minor issue with the DVB, even though it WILL WORK and is a more conservative calculation, as long as a potential Bot Operator is aware and does not fall into the same circumstances that befell STAX, apparently...
You can calculate the Daily Voting Budget by using the formula Daily Voting Budget = 100,000 / Target Voting Power. So with a 90% TVP you have a 1111% DVB.
It seems this formula is an approximation rather than a direct measurement. I took direct readings of "VP Expended" when using a series of 100% up-votes. The progression is not Linear, but an Asymptote. Meaning a "number trending to zero but never equaling it". So your first 100% UV is the largest, your next is a bit smaller, and the third is a tiny bit smaller yet. As far as the upper ranges of the VP goes, this will be "close enough" and actually conserve the VP a little better than a more exact calculation. The number my method gave was a bit higher than 1111, (1123 iirc) and even more of a gap above the number Phil told me at one time. In fact, the March Spread Sheet shows 1162.5 being the figure used, except it does not include the self votes or the swaps. Link to the March sheet:
https://docs.google.com/spreadsheets/d/1X7gkutia1lnBA44NS-IhEwNpPj533Jq8iqF6vpS7vuE/edit#gid=0
In my experience with my main bot @minnowbootcamp, I calculate my DVB with a flat 1100% figure and I use the "extra vp" to reward those that interact with bot's update and policy posts. I also have a MOTW (minnow of the week) award that nets the winner a 100% UV as a bonus. I suspect in SSG we will just want to figure it closer than that, and not have "excess VP to Burn Off" so to speak. But there are ways to do it, which should be in writing to avoid conflicts...
Very, VERY Well Done, @buggedout! Regardless of minute differences in figures, it will work. It might just generate a bit more "excess" than we expect, and then again ALL these formulae "assume 100% posting participation" so there will always be excess VP no matter what. Going upwards to a DVB of 1200 or more is dangerous ground, because the dynamic of a large community and varying posting habits can break the bank in a single day. It might work well for a week or so, then BAM!
I know these things are obvious to you, I am stating them for those less experienced in BOTTING.
- undie
I call it the Daily Vote Budget because it is intended to be a Budget. Most businesses and households ACTUAL spending can fluctuate either side of the Budget and it’s just supposed to be a guide. It’s not an exact science.
I trust the design of the system. When doing a 100% vote the formula is New Voting Power = Old Voting Power * 0.98. It is actually designed to be non-linear and it ensures that any vested SP can only dish out a certain fixed amount of rewards over time. It’s actually quite brilliant and I may write a post just on this topic alone because it is so poorly understood by many Steemians.
The Target Voting Power could be 50% and the community would not lose a single cent because the formula adjusts the value of a 100% vote and you can make more of them. A TVP of 50% is actually easier to manage because you have more room for error and it can fluctuate. As long as VP never hits 100% there is never wastage. The reason why many like to keep a high VP is to preserve the absolute value of a 100% vote but if you’re spreading it amongst 100+ people and not trying to keep a 100% vote for yourself then it really doesn’t matter. An equilibrium will be found for votes cast at any VP.
Anyway, I am aware that some users will post daily and others will not and this needs to be accounted for in the budget as it can cause a surplus of VP (or "excess" as you put it). I can get access to SteemSQL and profile the voting behaviour of the community as I have done in the past. I think I can manage that aspect with a bit of room for error and tolerance from the community that I’m doing by best. Nobody can really predict how many posts will be made by community members on any given day.
Excellent Grasp of the mechanics of the system, Bug! I've actually learned quite a bit here myself.
I was aware of the range thing, you can make the target anything you want. I have read some fine posts on that exact thing, but the writer was actually just using anecdotal evidence rather than the actual facts as provided by Steemit Inc. Is all of this found in the white paper? I suspect it is but I have not had time to dig thru it in that much depth. Good Stuff!
I do like the Target at 90%, that being the median of maximum vote weight for the recipients. I think most SSG Folks will too ;)
Thanks for the detailed reply.
I don't think it's detailed in the White Paper. It's a specific thing that was introduced in Hard Fork 19. I did a spreadsheet a while back to prove it and understand it myself, but if you want the documentation look for details of the voting changes made in HF19.