@Lembik Hah, I love suffix trees too. It was the first lame idea I initially had though, so I thought I'd put it up anyway :P

I think the interesting part about the lego one is if we require printing an error message if a rotate operation makes it collide with itself

Since chat doesn't seem to have much activity, suppose we went with the lego idea, where people have to implement a data structure that models a lego structure. What would be the winning criteria?

Off the top of my head, the first thing I can think of is fastest-algorithm

the first thing I can think of is always code golf :P

It's a fallback criteria, but personally unless it's just implement X I'd prefer to avoid it for this one :)

(I think the only reason we don't have many data structures problems is because half the questions are code-golf, making them usually a bad fit)

Code golf as a winning criterion makes me assume the winner will be one of the worst data structures by any other criterion.

@trichoplax the appeal of code golf is that no existing data structures are optimised towards golfability

Yes it certainly gives more chance of working on something there isn't already an example of...

(my comment wasn't meant as an argument either way - just thinking aloud...)

I also would prefer fastest algorithm, smallest number of bits used or some combination

fwiw :) (not much)

We could also choose something related to the structure itself, like "lowest proportion of the structure touched/changed/accessed during the operation"

For example if the structure needs rearranging as part of inserting or deleting

I've added voting comments for the individual suggestions in the sandbox post

Hopefully this will help us narrow down so we don't run out of time