What's up with the SilentBot curation 'upto' x% bit for promised votes?

If you are reading this, chances are you read one of the SilentBot star rating texts, and read that you post would soon be upvotes with uptu some percentage.

In this post I'll try to explain what this upto bit means, and how it gets determined.

If you look at the feed of the @silentbot account, you will see an hourly updated daily status report of the bot. At the bottom of that report you will see a table labled as Voting backlog. There is a number of voting accounts listed, but the important one to look at is the @silentbot account.

In this example, @silentbot has a voting backlog of 15 posts with a total weight equating 4.5 times a 100% upvote. In the last column we see the effective backlog. Basically this value, in this case 23 hours, seven minutes and a bit is the time it will take, given the current voting strength ang backlog for the backlog to empty and the voting strength to potentially reach 100%. But in this 23 hours and then some time new content will get curated and the voting backlog will get filled at one side while it's emptied on the other side. If nothing was done, chances are the effective backlog will keep growing and growing until post would start falling out of their upvote window inside the voting backlog.

Hope it is obvious that we don't want that. We need a fix, so this can't happen. The fix we are using is vote attenuation. As long as the effective backlog stays below one full day, votes aren't attenuated in any way. As soon though as the one-day threshold is reached, we introduce a linear increasing attenuation percentage that totals up to 15% per whole day. So for example when the effective backlog reaches one day and one hour and next item in the voting backlog is supposed to get a 90% upvote, there will be a 15%/24 is 0.625% attenuation of the 50% upvote, meaning that 0.625% of 90% will get deducted from the 90% vote, giving the post an upvote of approximately 89.44%. If the effective backlog continues growing to say four whole days though because there is that much top content getting posted and curated, there will be a 15%*3 is 45% attenuation of the vote. So at a four-day effective backlog, a 90% intended upvote will end resulting in an only 49.5% upvote from @silentbot.

This strategy means the larger the effective backlog grows, the smaller the actual votes become, and the quicker the actual backlog can shrink again. The system will find its own equilibrium, and I can't predict where that may lay.