The randomness of the multi-start puzzles drives me crazy. I want to work with particular starts, which is not easy in the UI.

This script solves that problem by restarting 100 times, and finding all the unique starting scores. It saves the first three of the starts to the first three quicksave slots. There are actually 10 quicksave slots, but the other seven can only be retrieved through code. If you change the parameter at the end of the script, you can retrieve any of the first 9 starting points. It defaults to loading the start in quicksave slot #1.

You may need to hit the 'Home' key to center the start into the window after you run this.

It reports information on the starts to the output window.

If you call the function with no arguments, or pass in 10 as the argument, it will restore the original structure when the script was run.

Very useful!

The structure that you have before running the script is saved to the 'recent best' position, so that if you go to the Undo menu and select 'Restore Rcnt Best' you will get back where you started.

It also sorts the scores, so the start with the highest starting score will be in slot 1, the second highest in slot 2, the third highest in slot 3. Lower scoring starts will require changing the argument to SelectMultiStart [e.g. SelectMultiStart(4)].

It still does not use quicksave slot 10, so if the algorithm is incorporated into other scripts, they can use quicksave slot 10 without fear of stomping on one of the quicksaves created by this script.

Also useful in other scripts: if you call the SelectMultiStart() with no arguments, it returns the number of starts.

..this SelectMultiStart() function should be implemented as standard game function!

This script doesn't help me. I ran it against 497 (Jan 2012) and it just looped through 100 starts in 50 or 60 seconds and they were all hugely negative. I think I tried it with another multi-start puzzle a few weeks ago and had the same issue. Rav3n pl SBS seems to handle 497 much better. I don't know enough about scripting to explain why. Not even sure that starting with a higher total from the first script will give me a higher total after a couple of rounds of tuning... but that's my feedback on this particular script. The other tlaloc scripts I have downloaded are *waaaaay* more helpful than this one.

If anyone knows how to use this better please speak...