« first day (1297 days earlier)      last day (3546 days later) » 

12:42 AM
@Sparr I don't think that algorithm will work
It doesn't allow for multiple people offering the same thing
I think I'm going to go back to the original way I have it
especially because that more closely aligns real life
 
@NathanMerrill err, it definitely allows for multiple people offering the same thing, you just treat a 'want' for X as a 'want' for every X
 
no, multiple offerings
Person 1 offers product A, Person 2 offers product A
 
1:12 AM
if person 3 says they will trade for A, that's a "want" for 1's A and another "want" for 2's A
 
2:08 AM
so when I say "I'll trade 3X for 4Y" that should make a link from my send-3X to anyone who is willing to receive 3-or-fewer X, and a link from anyone sending 4-or-more Y to my receive-4Y
 
2:59 AM
if implementation is all that's stopping you from using it, I'd be up for trying my hand at augmenting the edge generator to handle the cases we've discussed
 
3:29 AM
0
A: Proposed Question Sandbox - Mark XIV

Nathan MerrillTetris Wars In this KoTH, you will be playing Tetris with another player on a standard grid size. Completed lines will not be removed. A player gets a point by having more pieces than the other on a completed line. You win a game by having more points than your opponent. You will play each o...

 
bah, ninja'd again by the both
I was just going to post that
bot*
 
why don't you want people to rotate the pieces?
 
3:46 AM
I can't figure out a way to make it simple
aka, say I rotate a t-piece left. What happens?
also, it makes it easier to try to compete against your opponent
when he has a limited amount of options
 
sorry, I had wandered off
When you rotate a piece, it should always be in the upper left as far as possible
so:
____
|XXX
| X
|

____
| X
|XX
| X

____
| X
|XXX
|

____
|X
|XX
|X
 
4:03 AM
bottom left I think
that way it's as close as possible to 0,0
ping @Sparr @MartinBüttner what do you guys think about allowing rotating: meta.codegolf.stackexchange.com/questions/1847/…
 
it's barely tetris if there's not rotation
 
Ok. Rotation added
 
make sure rotation in tight spaces works the way it's supposed to
oh
you're only allowing drops
not rotation mid-play
I guess that makes my admonition moot
 
When I said top-left, I was thinking of the top-left cell as being 0,0
Also, I don't think you want the bottom to be your reference pre-drop
It's confusing in the tetris context
 
4:19 AM
mmm
you are right
I didn't think about that
Would you enjoy coding a bot for it?
 
I think bots will tend to block lines they can't win
if you have 6 squares in a line, why would I finish it? I'd rather drop a piece that makes it un-finishable
heck, even if you have 5 or 4, I can block "your" line and get started on the next line up
 
You are right. Score every line then?
 
I suspect that as written, the typical game would go 0:0
maybe score every line. maybe give a bonus for finishing a line. bonus of 1-3 points might offset the desire to block
but you'd know if your opponent could finish, so that doesn't mean much
score every line would make a tie game. both players would score every square they play, 4 per turn.
 
The only bot I can think of is to score the board, look at the score for each of my moves, and make the highest-scoring move
If I can see the other player's next piece, I do the same but min-max
 
you can see the next 6 pieces
 
4:26 AM
you can see their next 3, I think
a shallow enough tree to min-max the whole thing, I think
 
well, it's pretty easy to build a tree
yeah, it's only 40^n
before pruning
and that's only for pieces with 4 distinct rotations, and it's not quite 40 even then
 
Also, scoring every line wouldn't make for a tie game
If I win by having 8 pieces, I'm worse off than by winning by 6 pieces
 
so let's say (30)^n on average. You'll still get people playing perfect games
 
Hm. you are right. I don't know of a way to make it not min-max-able
except, I did have an idea to make everybody play on the same board. But I think that would be too unpredictable to make a good bot
 
You could make the pieces random and not visible ahead of time, but that doesn't really help much. Then people just max their current turn
 
4:31 AM
that would make it worse, I think
 
yes, probably, but it's not strictly worse; it does have the advantage that you have to speculate if you want to take the other player's moves into account
so in theory, a bot that tries to strike a balance between maxing its own score and screwing the other player could beat a bot that just maxes its own score
 
would only allowing rotation of your opponents pieces help?
 
not clearing lines is the problem. with line clearing, the strategy gets a lot deeper, for dealing with partially full lines below the current line
 
With a lot of pieces visible, I'm fairly certain you'll get optimal play, at least for several moves in advance
Oh, plus: with line-clearing, making the game end when you're ahead is a strategy
It's another component of the strategy that could make bots a lot more interesting
 
that's still a feasible strategy with no line clearing
 
4:34 AM
You could make the game end either when the board fills up, or someone reaches some score limit
 
but I like your point @Sparr
 
@NathanMerrill elaborate on scoring every line not being a tie game? every turn each player gets 4 points.
 
You get 1 point for 1 line
not 1 point for 1 square
 
ahh, got it
so, once you have 5, a line is a lost cause for me?
 
basically
 
4:36 AM
write an interactive version, play against a random bot. see how the game goes
 
which is why I wanted blocking to be a valid strategy
but I see why I need to score every line
 
hell, play against yourself, on paper :p
 
no, play against yourself using legos
Do this right
 
hehe
Ok. Lines get removed and you get to decide what your next piece is going to be (but not your current piece)
I think that should be sufficiently un-min-maxable
I'm still not sure whether uncompleted lines should be counted
 
that doesn't make it much less min-max-able
 
4:45 AM
there are now 6 times the amount of possibilities at each step
 
I don't understand the decision to let the player pick his next piece
That means he gets to pick every piece except his first one?
 
yeah, but its delayed
I could code my bot to say "find the piece that will complete that line"
but then I could also code my bot to say "find the piece that will stop what I think the other bot is going to choose to complete the line"
bah, that's still min-max able
 
yeah, I really don't like the idea of choosing your own pieces
 
I'm working on ideas to prune the move search space for the horror movie search challenge
 
bah, Deleting the post
 
4:52 AM
Cool. I really want to see alternate solutions. If you run Martin's solution with the graphical controller, you can see some really suboptimal behavior
He did say ahead of time that he was going to make a simple solution, though, so I expected that. However, you can probably beat him
One thing he does that I didn't expect to see is to leave one or two squares unsearched, go all the way across the map, and come back to sweep them up at the end
 
I need to get familiar with some graph search algorithms and/or libraries.
I think I have a strategy.
not the best. probably better than Martin's.
 
5:09 AM
Yeah, though I don't think any stock algorithms are directly applicable; that was the main point of having a sight radius, multiple people, and the death rules
 
oh, sure. I'm just thinking simple BFS
 
I asked my facebook (which includes a lot of strong females) if it's misogynistic to mock Microsoft by writing 'Ms Windows'. ... And no red-flags! .... ccuz iz a wee girly Oh Esss.
 
I look forward to checking it out
 
sigh. non sequitor.
 
@EricTressler is there a reason my script needs an updated map from the controller?
isn't the outcome of my move deterministic? I can just spit out a complete solution based on the map input
 
5:14 AM
No, you don't need to use the controller
Martin suggested I include it in case people were too lazy to handle the rules themselves
 
ahh
 
If your code conforms to the spec, the only difference in using the controller and not using it is that you'll get some stuff hitting your program's STDIN, and you'll have a chance to see the graphical output
So you can either not use it at all, use it and ignore its input just for the visual, or rely on it to enforce the death rules and deal with the sight radius for you
 
5:41 AM
10 lines of very high level pseudocode written for my algorithm
now to drill down :)
 
the high level is the hardest part, anyway
 
will be when there's more competition
my high level algo is still pretty simple
 
6:09 AM
One more day until my bounty on dogfighting runs out, I hope some more competition shows up
 
7:06 AM
@Sparr I added Penrose tiling to my framework. Still working on making it expand on demand. I'm using the pentagrid construction method, so in principle it should be simple, but I've obviously failed to understand something.
p3 seems to be fairly rare. I excluded periods 2,5,7,2*5,12,2*7 and ran overnight and got 19x p3, 108x p4, 41x p6 (although not sure yet how many are 2*3) and a few multiples.
Also 35 which hadn't stabilised in 400 gens. Penrose seems to be the closest I've seen to far to the original square grid in terms of the balance between everything stabilising quickly (most of them) and almost everything being unstable (rhombille).
 
 
2 hours later…
8:58 AM
@NathanMerrill How come you deleted the Tetris proposal? (don't have time to read the transcript)
@Sparr I suppose it's really simple to improve my submission just by including the field of vision in the BFS
 
9:44 AM
0
Q: are pure kolmogorov complexity questions appropriate?

Nate KerkhofsI currently have a few ideas for questions involving kolmogorov complexity, mostly how to output song lyrics in the shortest method (example: Diggy Diggy Hole by the Yogscast, http://yogscast.wikia.com/wiki/Diggy_Diggy_Hole_(Song), mainly because I've had it stuck in my head since the day of rele...

 
 
2 hours later…
11:19 AM
Could someone have a final look at this proposal? Especially if anything is unclear of if the post can be shortened? I'll be posting in 30 - 60 minutes.
2
A: Proposed Question Sandbox - Mark XIV

Martin BüttnerDomino Circuits code-challenge logic-gates It is possible to build simple logic gates from dominoes. Hence, by combining these or otherwise, arbitrary binary functions can be computed with dominoes. But of course, everyone who has played with dominoes (except Robin Paul Weijers) has experience...

 
0
Q: How long should I wait before accepting an answer?

TanisSuppose I write a code-golf question. How long should I wait before accepting the shortest answer? One week? Two days? More? I checked the FAQ, but couldn't find any information on this topic. Is there a general rule or is it just when the question asker feels like it?

 
11:33 AM
@NathanMerrill wait, you even pinged me during that discussion?
 
@MartinBüttner there are "scores" against the "non-optimally solved examples"
those might want re-naming to avoid confusion
 
11:54 AM
downloaded MCell, was not disappointed
Larger than Life is awesome
 
@MartinBüttner all sounds good - the only minor ambiguity is the following:
I'd change
"No circuit must contain"
to
"No circuit is permitted to contain"
I think everyone would know what it means anyway, it's just technically someone could decide to interpret it as there is no requirement for more, rather than there is a requirement for less
@MartinBüttner as to the entry barrier, I like the inclusion of example logical gates so that a valid entry can be made simple by combining these. This allows people to make a working entry fairly easily, and then focus on the interesting bit...
 
@MartinBüttner It might also be useful to point people into a direction as to how to create a boolean function using standard logical gates. That, along with the examples will makje for an easy simple solution.
Otherwise, it looks good
 
12:17 PM
0
A: Proposed Question Sandbox - Mark XIV

NarmerJenga code I think everyone knows what Jenga is. As per Wikipedia: Jenga is a game of physical and mental skill [...] During the game, players take turns removing one block at a time from a tower constructed of 54 blocks. Each block removed is then balanced on top of the tower, creating a ...

 
@overactor that's an excellent idea
 
12:28 PM
Thanks for the suggestions!
@overactor to what extend. I think if I point them towards Karnaugh maps that might already take too much of the fun away.
maybe just point them towards CNF or DNF?
 
so i tried out inventing my own CA (with arbitrary states!) and here's a sample run of a little simulator program I wrote:
X dimension of board: 10
Y: 10
Rule: 3,5,6
0 0 3
-1 -1 4
1 1 4
-1 1 5
1 -1 5

0000000000
0000000000
0000000000
0000000000
0000405000
0000030000
0000504000
0000000000
0000000000
0000000000

00000000000000000000
00000000000000000000
00000000000000000000
00000000000905050000
00000000071208050000
00000009122112090000
00000005081207000000
00000005050900000000
00000000000000000000
00000000000000000000

00000000000000000000
00000000000000000000
00000000090000100500
00000000000944231000
00000900701286440000
i didn't think it'd grow that quickly xD
X dimension of board: 9
Y: 9
Rule: 3,4,7
0 0 4
-2 -2 5
2 -2 5
2 2 1
-2 2 1

000000000
000000000
005000500
000000000
000040000
000000000
001000100
000000000
000000000

000000000
000000000
005000500
000949000
000444000
000040000
001000100
000000000
000000000

000000000000000000
000000000000000000
000014180018140000
000018093409180000
000000250425000000
000000001600000000
000001000400010000
000000000000000000
000000000000000000

000000000000000000000000000
000014032032036032032014000
000032059093088093059032000
 
12:45 PM
@MartinBüttner sorry about deleting the post.
I thought you were active at the time
 
don't be sorry, I was just wondering why it didn't work out ^^
(since I was thinking of doing a Tetris code-challenge myself)
 
because it was difficult to not be min-max-able
 
ah okay
is that due to the lines not vanishing?
that is, would an actual Tetris challenge not suffer from the same problems?
 
Well, no, I decided to make them vanish, and it is still min-max-able
 
what sort of search space are we looking at though?
 
12:50 PM
@NathanMerrill are the tetrominoes deterministic or how would you plan ahead with a min-max approach?
 
I played with both of those variables
 
btw my catch would be to make it real-time-ish. that is the time limit for each move would decrease every 10 vanished rows just as in real tetris... so even if there is an optimal algorithm it would still be a decent fastest-code challenge.
 
hehe
I wouldn't have that
if you can see ahead, it is easier to min max
 
I'll write that up soon, I think
 
but it is a random piece, it is just 6 times the number of options
still min-max able
 
12:55 PM
that would blow up much more quickly though, right?
0
Q: Domino Circuits

Martin BüttnerIt is possible to build simple logic gates from dominoes. Hence, by combining these or otherwise, arbitrary binary functions can be computed with dominoes. But of course, everyone who has played with dominoes (except Robin Paul Weijers) has experienced the disappointment when running out of them...

 
@MartinBüttner Does this mean you have time to program plane AI now?
 
@overactor hmmmm. 23 hours for the bounty... I'm not sure I'll manage to do it today.
 
@MartinBüttner Did you see the new helper function I added?
 
also, I had another look at it yesterday, and I don't think I can get my head around it well enough to be any competition for the existing submissions
 
@MartinBüttner There's only 1 decent entry at the moment tbh
@MartinBüttner I like that catch, as I've said before
 
12:59 PM
It probably wasn't even my idea :D
 
@MartinBüttner If you have time, you should post in my Traders KoTH :P
it'll take less time than the other ones
 
my Domino Circuits code doesn't fit in 3k characters ;)
maybe I should remove 3 out of 4 indenting spaces...
 
rather
 
@VisualMelon btw I've added more test cases yesterday
they're on bigger end of the spectrum though
 
1:03 PM
oh no, more opportunity for my program to timeout
I don't think I'm going to be able to remove 8000 characters from this
 
whoa, is it that big?
 
@MartinBüttner Someone asked me that last night
 
@MartinBüttner is there anything in particular you're struggling with understanding btw, I might have to update the question to make things clearer.
 
@Rainbolt I knew I could summon you :P
@overactor No, I think I know how the game works, it's just that I haven't really spent a lot of time with thinking about how to play it well.
 
1:05 PM
I'm seeing my infinite-growth-or-glider detection code tricked by the Penrose tiling. p2 * p5 * p7 * p19 isn't detected as an oscillator within 1000 gens.
 
@Rainbolt I'm disappointed you didn't say "your mom..."
 
@overactor That would be a rather strange thing for me to say, if you knew me.
 
@Rainbolt never claimed I knew you
 
@overactor Never claimed it wasn't strange (why do people do this)
 
@VisualMelon I guess you could always provide a link to a gist or pastebin and use your post for some explanation of the algorithm and a smaller and a larger example circuit.
 
1:07 PM
@Rainbolt Now I don't know what we're saying anymore
 
Then again it will be might be very disheartening to others if they see that the only existing submission doesn't even fit in an SE post. :D
 
it seems to fit if I remove all the indenting ;)
 
@overactor My job is done here.
 
The core logic should really be in the post, even if some of the support stuff doesn't fit.
 
... there is alot of core logic
 
1:09 PM
@Rainbolt Done, my job here is.
 
Someone needs to post a multi-challenge riddle, leaving bits and pieces among many other challenges over a few months, and then posting the signpost that says "start here"
 
@Rainbolt That would be amazing.
 
What was that website that has hints in the javascript comments?
I think it would be cool to mimic that in PPCG posts.
 
loads of them do
 
Where the hints are hidden in the edits
 
1:12 PM
my favourite one is ouverture-facile.com
but there was one by some PPCG user
 
^ that one
The only problem with this is that an unknowing mod could come by and wipe out a starred comment that was holding information
 
Joe Z made it
 
You could leave hints in deleted answers with "Take a break until you are well known." in the hint.
 
@MartinBüttner I feel so stupid because I'm still stuck at level 4...
 
1:15 PM
I think I got to the astronomy room and then couldn't be bothered any more
 
I got to room 7 or 8 and was completely stumped
I think it was the one after the dark room
No wait I got stuck IN the dark room
 
yeah same here
although I have an idea for it now
but I can't be bothered to get there again
and then I'd probably just be stuck on the next level :D
 
I Googled every single one of those celestial bodies, wrote down their initials, tried to figure out if their position in the tic tac toe board was relevant
Like if top left corresponded to the other top left
I really like those riddles because you have to have an "Oh, I see it." moment to progress
 
I think I got to the last level. Spent ages on it, and still not sure whether I got the right star but didn't use the format of the name which was expected, whether I messed up the maths, or whether Joe messed up the maths.
 
@Rainbolt IIRC they were asteroids so if I gave it another go I'd look at their official number
 
1:21 PM
@PeterTaylor your numbers for p3 p4 p6 seem to match mine. I also find a lot of 5 and 7, enough so that I've had a few false positives for p35. Not stabilising in a grid with dead border is common enough, as well. If you get growth working, I'd love to see it!
 
@MartinBüttner I remember looking at their number, but after quickly seeing that they weren't adjacent I moved on. Maybe there is a non-adjacent pattern
 
@MartinBüttner my actual approach will be different from yours, if I get it working, but one other change you might make is getting your groups to walk with 1 or 2 spaces between them.
 
Once I'd suppressed those I got a lot of p11 and p19. Currently my "is interesting oscillation period" logic is: period > 7 && (period % 5 != 0 || period % 25 == 0) && (period % 7 != 0 || period % 49 == 0) && (period % 11 != 0 || period % 121 == 0) && period != 12 && (period % 19 != 0)
 
oooh, p11 and p19. I didn't get any of those.
 
Turns out the bug in my optimised tiling class was due to not copy-cloning the vertex data I return to the renderer. Oops.
 
1:24 PM
@Sparr, convince @MartinBüttner to take on your planes
 
lol
@MartinBüttner the dogfight challenge is seriously lacking in competitors. you should come kick my butt over there :)
 
@Sparr thought about that as well, but couldn't be bothered to write code for coordinating them in that case
 
@MartinBüttner move "the leader", BFS a path from the others in the group to the appropriate distance from the leader of their group?
worst case, they follow along, but they would slowly spread out too
anyway, just an idea
I'm still failing on my "don't let people die" logic
without explicit groupings
 
My idea to improve the BFS (and remove the randomness) would be to only search until a move uncovers new cells (which would also speed up the BFS a bit) and then when I select a move (where ties are broken based on number of cells uncovered), then I mark those as uncovered on the map before doing the BFS for the next group.
 
that's my vague plan, as well
just with additional logic to avoid anyone dying
 
1:30 PM
oh guys... would anyone be so kind to consider giving my answer here its 15th upvote: codegolf.stackexchange.com/a/30286/8478 :)
(I usually don't come begging, but I have a good reason this time :D)
 
@MartinBüttner done
cool answer
 
thanks :)
 
@MartinBüttner I'm curious now. What was the reason?
 
I can hopefully let you know in a few minutes
 
@MartinBüttner Let me guess... Generalist badge?
 
1:38 PM
@ProgramFOX yes. but I just noticed that they're not given out until the 40 top tags have at least 200 questions each o.O
 
@MartinBüttner How far off are we?
 
very far
 
Yes, I just saw that too on MSE. I think it will take some time until we get that.
 
the least-used of the top 40 tags have 20 questions
well then thanks for the two upvotes anyway :D
 
We'll just post a few thousand questions tonight
Do the questions still count if they get closed?
 
1:43 PM
yeah it's definitely not happening yet, or Howard and primo would already have one
 
Where's @Calvinshobbies when you need him?
 
:D
well so much for that... back to hunting tag badges then :D
 
or chasing bounties?
 
@MartinBüttner Where does it say that?
 
1:49 PM
here is the deal: I'll write a plane for you if you write a domino circuit solver for me :P
 
@MartinBüttner Okay, I'll try my best
 
I've had a look at Sparr's wrapper... looks convenient enough for a Ruby submission. I'll see what I can do, but it might be tomorrow night rather than tonight.
 
MASSIVE answer posted
 
I would recommend using java, because of the helper classes/functions
 
@VisualMelon awesome :)
 
1:55 PM
but it would be cool to have a competitive non-java submission
and once you have a basic plane, the difference vanishes
 
@overactor speaking of which, do I need some IDE to compile your Java code without trouble or can I just feed everything to javac?
 
@MartinBüttner Should be fine, @Sparr did it without an IDE I think
 
need to add a couple of things to my answer before it's conformant
 
might as well do it in Java if I have to figure out compiling java anyway
 
1:57 PM
javac -d . *.java
java Planes/Controller (iirc)
so much easier from the command line than in Eclipse, imho
 
In eclipse, I just click a button and it runs :D
It's a nice green button too
 
@overactor sure, and when you want to run a script to parse the output after each run?
 
@VisualMelon the GIF doesn't seem to be looping
 

« first day (1297 days earlier)      last day (3546 days later) »