Illegal Position Contest!

Helpful factoid, 6 pawns from both sides can pass each other and promote without involving the capture of other pieces.

Would probably lose my mind though trying to calculate exact promotion squares for that last one lol!

I've just realised one possible retraction that exists, starting from Leither's diagram after 1... Rcb2+. The retraction Re4-e8+ is possible instead of Rc2-b2+, but that can easily be fixed by replacing Re8 with Qe8 in the initial diagram.

Hopefully I've made this position robust.

Nice one! I imagine it was quite difficult selecting positions for each piece; there's always so much to keep track of.

Last moves could be:

White needs 7 pawn captures. Black has PPRRNNQ available. wK passed thru c7 to reach the back rank. bPg6 captured an N. Seems legal.

You're right, I intentionally left d4 open to allow for a discovery but unfortunately forgot that the bishop can fulfill that role.

The missing white h-pawn allows for your discovery to work:

Here's a second version thats based on a similar but slightly different reason. I haven't really checked this one either but see if this works any better.

I guess the knight discovery can be fixed by just replacing a rook with a knight (which is actually what I originally had... maybe I shouldn't change things without checking them)

I was hoping it would still work with two rooks to be more tricky, but I guess I just tricked myself. The rook discovery for LSB was also a later addition that I thought would be handled by the ep, except I forgot the ep isn't forced if I introduce a discovery.

Hmm, but now there's a spare black unit to uncapture? [No more bB on c8]

-1.gf6ep f7f5 -2.g4g5 Kd6e6 -3.Qxd4+ etc.

i suck at making good puzzle

Legal. Last few moves could be:

Both sides have a promoted Bishop, but that's fine.

Yeah, the battery becomes a problem. I'll to see if I can try to make something interesting, because most of the easy fixes I can think of aren't really that fun.

Wow, this ended up getting really complicated, but hopefully this is both illegal and interesting to solve. If I am correct, there should be multiple cases, although some retraction sequences are much shorter. I spent a lot more time trying to find all the retractions and ended up discovering some really interesting retrograde challenges along the way, so let me know if you want to see those.

Let's just hope I didn't miss anything this time; most of the time I forget the reasons for certain safeguards that I do and end up removing them.

Seems legal. Last moves are:

White's f-pawn captured 4 times to promote to B on d8 or b8. One more P capture was on a3, and Black has QRRNN available.

Black shows 2 P captures; white has QR available.

I assumed Q promotion wasn't possible because white would need to promote to a rook, but it looks like I just got confused. Was hoping it would be more interesting that way but I guess I have to put the Q on h7.

-1.Nf5d6 d6d5 looks fine - wPf is on c6 and wPc promoted on b8/d8.

Edit: or even wPf captured to the c-file on light squares, then promoted.

Oh well, I was hoping K discovery would fail. I guess that means I have to put a pawn on that square.

You've probably already figured out the intended solution is for both R and K uncaptures to eventually fail, so I'm not going to bother reposting at this point. There's also a funny sequence if d5 is open where wR can do a staircase and do 2 discoveries in a row, which I thought was interesting. Of course you know I like my staircases :-)