First of all Miner 1 would be in trache 0 (because he only tithed 1 bbp). Meaning these two hashpowers would not be competing. Miner 1 can only "mine" in blocks with an interval of 0, (for example, block 80000) while miner 2 can only mine in block 80015 (interval 15). Miner 2 would certainly be competing in Tranche 15 because of the high tithe.
Next, lets change this up to this scenario:
Miner 1: hashpower 1 MH/s, tithed 1M BBP
Miner 2: hashpower 1 H/s, tithed 1M BBP
In this case both miners would be in tranche 15 competing against one another. Miner 1 would find the block but would be forced to pay the pool 80% of the winnings (deterministically enforced btw, by the wallet checkblock code). The only advantage Miner 1 would be getting for all that hashpower is the 20% bonus for finding the block.
Note that in this algorithm, miners who aren't in a tranche sleep during those blocks and wake up automatically when: Their tranche is called, the block is late, or our difficulty rises more than 100% over the historical reference point.
1. Ah, I got a little confused with tranches, I forgot that the miners in my example would be in different tranches. So if the miners sleep 15/16ths of the time, does that mean the CPU will be roughly at 0 for 94% of the time? That is certainly interesting and as the wiki says - green, but I wonder if there would be any economic or technical complexity because of that. By economic complexity I mean people who rent servers would have to pay for the entire server resource and use it for only 1.5 hours out of 24 hours in a day. By technical complexity I mean that the server providers could potentially automatically flag these spikes as something not normal and limit the server. But on the plus side, that will further improve the resilience and friendliness of the algorithm by making it harder to mine for "hashpower whales" and easier for home users, because the algo would be very friendly to their home CPU and their power bill. Now that I think of it, this seems to be a good feature, although very unique and untested so far, but I hope for the best.
2. In your example where they both tithe the same (or similar) amount and are therefore in the same tranche, if my math is correct, they would receive this many BBP per block:
Miner 1: hashpower 1 MH/s, tithed 1M BBP - 4000 + 2000 = 6000 BBP
Miner 2: hashpower 1 H/s, tithed 1M BBP - 4000 BBP
Hashpower is clearly not valued nearly as much as tithe, which I know was the goal. And yes, in a real scenario, there would be for example 100 users in a tranche, so each user would get 80 BBP when their block comes, and the block founder will get 2080, which is much more compared to others in the tranche (unlike the example above), BUT the chance is much lower, and it's not predictable how much you can earn due to the luck factor, whereas the 0 hash big tithers would have a much better predictability - the competition in a tranche would increase relatively steadily, like in sanctuaries. Also, you would only have 13 chances per day (205/16) to win the lottery of finding a block, so the luck factor is 16 times more unstable than when you have 205 chances per day.
Another thing it crossed my mind is that maybe a solution to the 0 hash big tithers would be something like having a requirement of producing a certain hashpower over the previous 24h to be able to tithe at all, and the higher amounts of tithe would be unlockable with larger hashpower.
3. If tranche levels are determined by arbitrary numbers, then I think the middle tranche levels would be the most empty. I want to explain this by the following example: let's say there is a poll named "How much do you earn per year?" and there are 5 possible answers:
a) 85k or less b) 85-90k c) 90-95k d) 95-100k e) 100k or more
The results would be something like this:
a) 87% b) 1% c) 1% d) 1% e) 10%
I put them close intentionally, but the point is that it's impossible to assess all people's salaries and evenly distribute the offered answers, or tranches. You would have to categorically know the statistical wealth of BiblePay users. Therefore I think the middle tranches would always be the most empty, even emptier than the highest one, because simply there are no further levels than 15, so all whales, big and small, would flock in that one tranche. Correspondingly, the first tranche or two will be a test bed for newcomers to try and mine, so those first few levels will always be busy. That means it will be more profitable to tithe less than level 15, to not be in the top tranche, and rather find another nearly empty tranche. So it could happen that the ones who tithe more would be rewarded less and vice versa. That doesn't seem intended. So maybe solve this by having incremental rewards per each tranche? So if we have 128k BBP for all 16 tranches (8k * 16), it could be distributed so that tranche 0 doesn't get 1/16th of 128k, but say 1/100, and tranche 15 gets like 1/5. But that also doesn't look fair somehow.
4. Could it happen that the earned BBP for a 24-hour period is less than tithed BBP? If so, that could put off many users, especially if it's very unpredictable what will happen. Or if the return could be roughly the same as the tithe. How can we ensure this doesn't happen? It looks like it would happen when the number of users increases, while the tranche levels and reward amounts stay the same.