Originally Posted by
Grumble
They don't solve to >95%. You do understand exactly how it works, and how algorithms work, right? I'm assuming you don't otherwise you would know exactly what it is you're proposing be done in a matter of hours.
Don't worry, when quantum processors get out of the lab and on the market you'll be able to get a PBS line optimized to around 99.9% in a few seconds.
No. The "solving to >95%" that you speak about is a result of only having 7 discrete bid groups. The solver assumes that all trips in the same weighting are equally desirable by you. So it goes up and down and does something called "brute force" crunching to "optimize" the lines. The problem is that if you had 100 weighting (i.e. 1-100) the solver wouldn't do this. It would optimize your line, and move on.
Most of what the solver is doing is wasting time while it moves trips in and out of your schedule that to the solver you are equally indifferent about. That is not true. If you were asked about those 5 trips all in the N bracket and asked to rank them from 1-5 you'd do it in a heartbeat. But you can't, because you only get 7 weightings. So if there are 35 trips you desire you have to bunch them in groups of 5. This means the solver can potentially "brute force" build you hundreds of combinations of lines. Doing this with over 10,000 pilots is why the solver does it.
So what we need is to let the PILOTS be the solver and give us 100 weighting (1-100) and then just give us the trips in that order. If we want to put all our trips at 50 weighting the solver can flip those in and out.
But the solver is doing busy work and not really accomplishing anything.
All the while we are told to just be a good little lamb and not question the system that's obviously flawed.