
What Went Well
Idea
It took me about half a day to come up with an idea I liked, but once I came up with this one, I was set on it. All of my earlier ideas were things that were funny to say like "tabletop + mmo" or "physics-based + text adventure", but none of them felt like something I wanted to make. I was feeling pretty dumb and getting close to giving up on entering this time around, but "match 3 + connect 4" managed to be both funny and strangely compelling.
Within a couple hours of having the idea, I was already well on my way to a functioning game:

Animation
One of the first things I mentally planned out was a small animation sequencer. I didn't end up building this until after getting the mechanics mostly working, so it took a bit of refactoring, but once it was in, adding polish was almost trivial! The snazzy intro, the explosions, pieces falling into empty spaces; all of this was driven by the same basic system. Having a solid plan for how this would work let me add tons of little details that would have taken much longer to implement otherwise.

User Testing
To be honest, I left this later than I should have, but doing it at all had a huge positive impact on the end result. Towards the end of the jam, I had a couple quick sessions with my sisters which led me to make crucial balance changes.
Before testing, the game was almost the same as the final product, with one key difference: there were no swap points. Players could swap pieces on their turn whenever they wanted. This conceptually seemed fine, but testing proved it was a terrible idea: it became far too easy to prevent your opponent from getting close to a win by swapping to make match-3s with their pieces and blow them up. The games inevitably became boring, perpetual stalemates.
I went through a couple quick iterations, but effective use of swaps always ended up dragging the game on. The first version which actually made a difference was the one in which I introduced points: in this version, every time a match was made, it gave that colour an extra move on their next turn. De-incentivizing blowing up the other player's pieces made a big difference: the board actually started filling up a bit, and complex strategies started emerging.
Ultimately, swaps still ended up being over-powered, and planning around extra turns made thinking ahead more difficult than I was comfortable with. In the final version, the players have to earn their swaps by matching pieces. This provided a great balance: defensively blowing up the other player's pieces is a trade-off between denial and giving them a powerful resource, and it also gives you incentive to blow up your own pieces in order to bank up swaps for later.
What Went Wrong
No AI
Putting up a local multiplayer game without an AI is always a big ask for LD. Some folks'll have ready access to a willing partner, sibling, friend, etc. to play with, but most people are just going to be able to demo it on their own. I had planned from the start to make an AI, but it was already submission hour by the time I had something kinda-sorta-but-not-really working. I think I made the right decision to scrap it, but should have budgeted my time better so more people could have a chance to play the game as intended.
Menu
A number of people have had issues with the default enter key behaviour in PICO-8. It's easy to take PICO-8 stuff for granted when you've been using it for so long, so this one kind of surprised me! I think this could have easily been fixed by changing the start menu to read "press z/x" or something along those lines.
Conclusion
Overall, I think this is one of my best LD entries to date. It's not perfect; there's definitely more I would've liked to do with it; but I feel like it was both a solid idea, and a solid execution of that idea.