scoring for non synthesised lab results is not correct

I think I’ve just confirmed that scoring for non synthesised lab results is not correct :slight_smile:

my alternate identity…

now has 12910 points - it had 10000 points
it voted for the winner - gaining 970 points (fine)

and created this monstrosity called “awful dotplot and meltplot” (which is nothing like the winner)…

and got 1940 points for that (perhaps it was so bad it was treated as being identical to the winner?) which is equivalent to a predicted 97% success rate.

Just thought I’d mention it - as it was so surprisingly unlike the winner it probably ought to have had very few points (which I gathered were earned for similarity to the winner)

Hi Edward,

since there is no good way to evaluate how good non synthesized lab design is, we evaluate it by assigning it the score of the closest synthesized design - in this case 97% design was closest to your design.

We realize that this is not the best way to score the non synthesized design, but currently we are not sure what other metric we’ll use.

Also, please note that this scoring metric will eventually go away when we have high throughput synthesis pipeline for 20000 synthesis per month.

thanks for the explanation - and I guess when the 20000 synthesise are being run - the old unsynthesised (non christmas tree) designs can be run and get the scores adjusted :slight_smile: