Thanks to those who chimed in with more data points. Knowing roughly the percentage of non-US applicants helps.
My post yesterday was meant for fun - I was playing with the numbers. I did make one mathematical error but the entire approach was superficial and it was meant to be.
So, yes, we do not know the distribution of runners' times but I think we can assume that we will see a larger density of runners per second in applicants as we get closer to the 0-5min applicants. Last year, applicants basically broke down as follows:
2019 Applicant distribution
A 17% - 20 mins+ better than BQ or 5256 applicants and 8.7 runners/min *
B 28% - 10-20 mins or 8620 applicants and 14.4 runners/min
C 28% - 5-10 mins or 8545 applicants and 28.5 runners/min
D 25% - 0-5 mins or 7604 applicants and 25.4 runners/min
*assumes that the majority of these runners would be clustered toward the 20-30 min range.
If we just play with last year's applicants as a proxy of the population of runners who (a) make the grade and (b) are interested in applying and we assume that this proxy holds from last year to this (it doesn't necessarily hold but this is just for sh*ts and giggles anyway).
AND
We assume that 25% of population A from 2019 will slip into population B for 2020 AND 50% of population B-2019 will slip to population C-2020 AND 100% of population C-2019 will slip into population D-2020 then this year's distribution could look like this:
2020 Forecasted Applicant distribution
A 20 mins+ 3942 runners
B 10-20 mins 5624 runners
C 5-10 mins 4310 runners
Total week 1 applicants: 13,876 runners
D 0-5 mins 8545 runners
And if the above held true, we would have 8765 spaces available for week 2 and 8545 runners vying for them (assuming the same 433 runner population for 10 consecutive BMs. And that's not what's happening according to update BAA registrations so far this year.
So... on to the goalpost effect
Has the toughening of standards drawn some runners from population C-2019 to improve their times by 5 mins and stay in the week 1 population? If even 20% of C-2019 achieved that outcome, then we could see another 1700 runners scattered across the week 1 population but significantly clustered in C-2020. If the number was as high as 25%, that would add a further 427 week 1 applicants. I am going to assume that the number will not be much higher than that. If it is around 20%, then there will be 7,000-ish spaces available for squeakers this week. That number could drop to as low as 6500 or even lower if the goalpost effect is stronger but my 20% goalpost assumption pegs it at 7,056 spaces for week 2.
Regardless, there are clearly thousands of spaces available for us squeakers this week. So, how many of us are there?
I expect the squeaker population will be larger than last year's but not by much - if a part of last year's 5-10 min population managed to stay in week 1, then they're not in the squeaker group this year and this will somewhat reduce the upward pressure on the squeaker group this year due to any goalpost effect from last year's squeaker group. If I had to pick a size of the D-2020 (squeaker) population, I would stick with 8500-ish - or 28 runners per second. Given that, I do expect there will have to be a cut-off and right now it looks like it will be around 47 seconds.
And, yes, I had fun crunching these numbers and I made TONS of assumptions - all of which are undoubtedly going to be wrong.