Extinct, a post mortem

Posted by (twitter: @@benjamin_soule_)
December 19th, 2012 6:49 pm

gorillaIn Extinct : Humankind fade in the mist, you play some kind of gigantic gorilla god who wants to get rid of humankind.

You have to put some traps on the way of a human tribe so they wont be able to reach the meat point.


The interesting part is that most of the traps leave the humans (AI) with an A/B choice. So you have to figure out a way where none of those choices lead them to victory.



Basically all I had to do for humans AI was testing all of the possibility including path and traps choices.

– If I find a solution I play it and player loose the level.

– If I can’t find a solution the player won the level but for me the troubles begins : I have to choose the “best” failure. The one in which you really think they tried so hard to reach the last point. Not easy when 95% of the failures looks like stupid random move. I implemented several algorythm without any success and finally decided to choose a failure where the human reach the most advanced point. But that’s not enough and most of the time, the failure sequence looks like humans just became dumb in comparaison with the win sequence.



In this kind of game leveldesign quickly become a nightmare. If you dont have computer assistance, it’s difficult to check if your level is working.

So I coded some kind of brain class who launch multiple solvers and tell me, for a level, how many combinations are victories and how many are failures.

Of course the best result would be  ( for a 4 point problems , 24 possibilities )  : 1 Humans Failure / 23 Humans Victories.

It means there’s only one solution to the puzzle which is basicaly your main goal as a puzzle designer.

Once it was coded the brain class was very handy. Maybe a bit too much, because leveldesign came down to randomly change parameters ( path, starting humans, traps ) until I reach 1-5% of humans failures.



I spend the rest of the time on traps animations. Each trap had to show very clearly what happens. I think the result is ok for the timing, but my sprites was too small. Some friends tried to warn me but it was too late : animation done and the levels were too big for going to X2 scale.


Only 1-2 hours left for music and SFX, once again sunvox and labchirp saved my life… But I wanted to try SION MML which was impossible in such a short time


I’m disapointed by the difficulty curve. I should have start with a 3 traps level instead of 4. It would have been  only 6 possibilities so people can finish it, even by accident.

My ingame help is insufficient, most of the people dont understand the fact that hero have more life points and the usage of spear or fire.


I know that puzzle games are not quite popular in LD because when you test 10-20 games in a day its very difficult to take time to solve brainfucking puzzles, so I think I really shoud have pay attention to this two points.

You can play Extinct here


3 Responses to “Extinct, a post mortem”

  1. loren says:

    I really liked this game, it was a lot of fun!

  2. Suese says:

    Outstanding game, but I found a bug. The volcano does not hurt the tribe leader, though it does not mention this.

  3. Suese says:

    Again Benajmin, you have made the most solid entry in the competition. I think you should win in overall.

Leave a Reply

You must be logged in to post a comment.

[cache: storing page]