I just created an excel table to help predict future development of the wr tt.
Download here
What I thought was this:
1. Finding new styles gets less and less probable
2. With time wrs get better and better down to a limit, the closer you get to the limit the slower the wrs will improve.
From this we can deduce that a good simple mathematical model is this formula (how this is done I shall not go into here

):
TT = A*exp(-bt) + C
This formula does not take into account random variations, like good people suddenly höyling like hell, good people getting bored, if some new genious suddenly appears or if a new elma version stops höyling... If the höyling doesn't stop but goes on like now this could still be a good model though.
What we can do now is estimate the parameters A, b and C to make the curve fit with the actual wr tt development. In my excel table this is made easier, all you have to do is this:
1. Fill in what you think is the best possible humanly reachable wr tt.
2. Chose two wr tables that the "predicted wr tt" should run through, and fill in the table numbers and total times (of those tables).
3. Redjust these values until you think the curve fits well compared to the actual tt development.
If you think many new styles are still to be found, you should make the predicted curve fit with the entire development of the wr tt, that is:
For example choose that the curve should fit exactly with table 1 and 189.
However, if you think most of the best styles have already been found you should probably make the curve fit only the last part of the wr tt development. For example choose the tables 160 and 180. (This is if you think only the last part of the wr tt development represents how it will be in the future)
One example of usage of this table:
I thought the first few tables didn't quite fit in (mega improvements the first weeks), so I chose to fit the curve to tables 4 and 189. Then I wanted to see if a wr tt of 35:59 seemed possible, so I filled in the value 35.99 (<-decimal format) in the limit field. It doesn's look that bad, does it? Have a look:
Graph 1 <-- Goes to table 189
Graph 2 <-- Goes to table 500
Sorry about the bmp's, was unable to save as jpeg here at school
If this mathematical model is good, and if the absolute limit is around 36.00, then this is how the wr tt will develop (on average) if people keep on höyling
If 35:59 is the absolute limit however, then this calculation shows that when approaching that limit improvements are really small. In the the last 100 tables (2 years) only a few seconds are improved. Probably höyling will stop long before we get that close to the absolute limit...