I'd like to request / propose a trivial yet impactful change with voting for posts.
The problem
"As everyone knows" voting power recharges at a pace of 20% per day. You can cast 10 full powered votes a day (I'll use only full powered votes in the text below), since each uses 1/50 of your voting power. There is a fine print though. It is true only if you cast your votes once every 144 minutes. That is because every time you use voting mana, your account temporarily becomes smaller. Upvoting is the only manabar that works this way. When you are using RC mana, you are consuming exactly the full cost of your transaction. When downvoting, as long as you keep upvote mana at full, all your 100% downvotes will be of the same power until you run out of downvote mana and start consuming upvote mana for downvotes. But when upvoting, the weight of your vote translates to power based on your current mana, not on your full power. If after one vote you don't wait for recharge, your next 100% vote won't actually be 100%.
Imagine driving an EV covered with solar panels. It recharges slowly on its own, but while you drive and battery charge is consumed, your max speed drops. Who would want to use such vehicle?
There are some consequences of such behavior:
- plankton accounts on the edge of dust votes have to always wait for recharge to not fall below dust threshold
- minnow accounts that can assure their single vote pays enough for the comment to actually receive reward are in similar position
- for every account: not only your HP and given weight of the vote but also order of voting influences its power; in particular, if you want your vote to be as strong as your stake allows, you have no option other than to wait for full manabar
All of the above can be avoided with use of bot. Instead of voting directly, just pass your chosen article to a bot and let it vote in your name once your mana is near full (or the voting window is about to close - whichever happens first). However I don't think users should be forced to use a bot to vote optimally.
There are consequences that cannot be avoided even with bots:
- large accounts, even if they spam votes every block, stabilize at some point, when their per-block-recharge is bigger than cost of vote - that makes some portion of their voting power unusable; it is actually true for all accounts, but for small ones their stabilization point might be below dust threshold (all accounts stabilize at 1/2880 of their full power assuming one vote every block)
- there are situations when there are more opportunities to vote on one day, while there is not enough on other days - for example #FungiFriday that I personally take part of; despite voting in four batches (Thursday night when first posts are published, then Friday noon and evening, finally Saturday morning for last posts), I still end up at 75% or less power, which means my "full powered votes" for last posts are only as strong as 75% votes on first posts, skewing the rewards away from authors that participated late in the event
Solution
The solution is very simple. Four new lines of code. Starting at next hardfork (1) use voter.get_effective_vesting_shares() instead of voter.voting_manabar.current_mana for calculating used_mana (2) - the same can be used for downvotes actually. The last line is in two copies - an assertion to stop use of mana the user does not have (3,4).
The change makes it so that:
- all your votes of the same weight will have the same power as long as your stake did not change in the meantime
- you can actually use all mana - some users might need to lower their vote weight in case they vote more than 10 times a day on average
- without any adjustments to vote weight you will have less excess mana to burn on hbdstabilizer
- you gain full freedom of when to use your mana within 5 day recharge window
That's all. Since the change is so trivial, even with tests it should not require too much time to add into code, therefore could still fit into HF28.
By the way, if anyone knows why upvotes work the way they do currently (@gtg?), I'd be very interested in that information. There might have been legitimate reasons to do it in such way and maybe those reasons still hold. Or maybe not, after all voting and reward calculation looked vastly different in early days of "legacy blockchain" (picked up this great term from @crimsonclad 😁).