12144. Tom 7 (c-67-186-7-254.hsd1.pa.comcast.net) –
05 Oct 2013 21:53:06
[ 39m33s or ⊥! ]
IT'S THE WORST |
| |
|
|
| |
|
|
| |
Only if it crossed 2 red lines on the cake. |
| |
You are my hero. :) The puck at the end is just icing on the cake. |
| |
Ice skates on pavement. Wow.
|
| |
12130. Franklin Chen (71.112.11.13) –
30 Sep 2013 00:03:45
[ 39m33s or ⊥! ]
I just stumbled across your blog, after seeing your hockey photo in the 2013 Great Race. I found your observations about the 10K distance to be interesting, because I have rarely maximized my 10K race performance. The one time I magically did what it took to really squeeze everything out was the 2002 Great Race 10K, where I did 42:49, which I never matched before or since. I had to train very, very hard to be able to mazimize the 10K, because it seems to require considerable physical and mental tolerance of a certain level of pain, which is a step beyond semi-pukey, and perfect pacing (because going too fast will cause major slowdown in the last mile). A very difficult race distance for me to do my best in. |
| |
I expect those are unicode emoji for "thumbs up" over and over that I just can't see on this system. |
| |
|
|
| |
"I think there are about 3 Tom font tattoos in the world. "
Pic or it didn't happen |
| |
cool new EP! still waiting on that "two-track short form old-days style" album promised. it's been a while... :( |
| |
Yeah like I said on the LD page, the theme makes this super hard. Any hints on how the spikes actually act at the end? It felt like they punch through anything I put up and I never had enough time to figure it out :< |
| |
Yeah!! I was going to volunteer to represent the connoisseur block but I can't even tell the difference. Last two are definitely my faves. Keep up the jamz! |
| |
Thanks! I did it by clicking on the computer. |
| |
Sorry that there's no tutorial. There is some explanation in escape.txt, I think.
The editor prevents you from placing a block on a panel, even though that's a legal state when playing the game. This was deliberate, to reduce the amount of hidden information in the game. Levels that have that or a couple other things will be filtered by the game into the "quarantine" section.
It is possible to get the same effect (more or less) by using broken bots, which mostly act like grey bots and can be placed on a panel in the level's initial state. These don't flip to the bizarro world, plus act like bots in some other ways (if you teleport atop one you die), but for pure Sokoban levels they suffice.
That said, I'd prefer if you didn't import a bunch of Sokoban levels wholesale, especially if you don't have permission from the original author. However, if you have some particularly good ones, and permission, then those might be good. Best is if you create new levels, especially if they are Escape-specific. :) |
| |
|
|
| |
I wish there's a better tutorial on the level editor.
I checked the existing IQ Carrier level (level 353) and tried to understand it from the level editor, but still not sure.
Specifically, I try to create a sokoban level that would have "box on goal".
So far, I think it only shows boxes and goals separately on the IQ Carrier level, it has no box on goal elements in there.
It would be nicer if the level editor allows importing XSB format of Sokoban levels. If it is already possible to do this, please let me know how to do it.
|
| |
You need to play kerbal space program. Also, learnfun is crashing on my computer and i'm having difficulty figuring out why. If you google my username I'm pretty easy to find agian. |
| |
If that's how it works, that makes sense.
Anyway, I think it would be interesting to see how playfun would work in a football (soccer) game. I'm sure there are some for the NES, and it might be cool to see how it decides its strategy. Not sure though if it would be able to understand the end result of scoring a goal.
Keep up the awesome work ;) |
| |
|
|
| |
It could be extended very far; if I had 10x the number of computers (or time) then 50 seconds. This wouldn't make it make better choices, though... the problem is that there are a lot more than 10x the number of 50 second input sequences than 5 second sequences (exponentially many!) so the likelihood of happening upon one that is good drops dramatically. I chose 5 seconds because it seemed reasonably close to the horizon that humans plan their movements under, when thinking about "go over there" plus the consequences. |
| |
Your research is very inspiring. :)
I read in the paper that playfun looks approximately 300-500 frames into the future in order to decide the best scenario to choose from. Did I get that right?
Anyway I'm very curious to know just how far do you think you can extend this, because the end result should be making the program "smarter" in the long run?
300 frames should be roughly 5 seconds, but if that were extended to perhaps 20 seconds (if that's possible) it should mean the program makes better choices, correct?
Thanks and sorry if I'm a noob at this! I'm just very interested in your program's capabilities. :) |
| |
Sorry... expect some rough edges; it's just research software! :) |
| |
|
|
| |
No.... I better try that....
(damn it Tom put it in the readme :(.) |
| |