I've been working on Res furtiva for last week and I'm happy to feel that the new version is much more pleasant to play. These are the highlights:

• lots of balance in enemy generation and stealth mechanics,
• lots of useful information is presented in action log,
• you can now visually “hear” enemy steps in vicinity, allowing you to plan your actions more,
• enemies have their patrol paths instead of just roaming around the camp.

Also many bugs have been fixed and there were loads of minor balance and UX improvements, including rewriting the intro in shorter but more interesting manner.

Much better message log. I had a sense of loss when removing “Step sound becomes aggressive!” bug. 

I was kind of aware of that in the beginning of the jam and it was a concious decision to push for anything I could imagine when I was thinking about the idea (I knew I have a strong framework and was prepared for working day and night). It was an interesting experience, but it's also tiring to just debug, clean and polish a project for whole second half of its life. Next year I'd rather check the reverse approach and go for something much simpler, yet complete to the last button.

And now precious design lessons:

I realised that some of the things related to mechanics cannot be polished or fixed in reasonable time. Morever, I should probably have resisted from spending time on them at all. For example — did you know that if you distract your opponent by shooting at something in his vicinity, he'll first stare at it, then come closer and look around? When I was implementing this, I imagined players using this realistic feature to attract the guards away, while watching them do this inspection ritual and being amazed by complexity of AI. There were two major problems with this.

• Firstly, games made during jams are rather not something that you'd like to experiment with. Looking at comments from playtesters, I noticed players think straightforward if this allows them to reach their goals (and I felt doing the same while playing other 7DRLs). I like current fluent player detection mechanics, but by making it like it is I let the players use simpler stealth technics (like sticking to walls).

• Secondly, even if a player chooses to attract an opponent away, he'll probably want to be at different place, so he won't see what exactly the guard does. This brilliant discovery makes me feel so stupid for constantly thinking about the gameplay from perspective of Unity scene where I can see everything.
  

Did it really help me?

One thing is sure now. Adding new features on top of current Res furtiva would be expensive. Mixing their own complexity with complexity of current state of the game (plus code refactoring) would result in hundreds, not dozens of hours, which I'll probably prefer to spend on other projects. But here's the list of the features if you're interested.

• time-management mechanics between missions; choose whether to train skills, recover from injuries, learn Gmetian language etc.
• story-related events between the missions,
• lockpicking through a mini-game based on Tetris and Sokoban mechanics,
• environment interaction including traps and lights,
• disguising as a Gmetist,
• new level types including manors, swamps, offices — actually this would be the easiest one on the list.

Remembering about planning most of these things for 7DRL version of the game makes me bitterly smile.  But to my defense, it wasn't like “mini-game on first day, three types of levels on second day”.  I know that during such jam you have to smuggle small scoops somewhere between core features, bugfixing and balancing, but I let myself (maybe a bit too much) spend more time in sphere of dreaming.