« first day (1268 days earlier)      last day (3754 days later) » 

4:00 PM
@Jon yes
profilers also help track down memory leaks
 
Jon
also shows managers some numbers they can show their boses
 
Pip
who downvotes?
-1
A: What coding issues can be significant in damaging the performance of a game?

PipIn my experience, there are some things that greatly affect the performance or my games, and some that do not, as much. Most Significant: Running lengthy code all of the time. This can be avoided, at least partly, by checking if any given code needs to be run that frame: if change == True: d...

 
Jon
What do you want, a list ?
 
Goddamnit, I want bacon now. And the best brunch restaurant in the city is a block away.
 
Pip
I didn't finish my post, and accidentally clicked the post button
 
Jon
i made a restaurant website in college
 
what's a segg? Why would I want a blue one?
 
Blue's Egg
l2read
they have awesome Wisconsin Bloody Mary specials.
 
Pip
can I has the upvotes?
 
and a huge full bar with amazing drinks, which I imbibe upon whenever I stop in.
 
4:03 PM
i only looked at the url and deliberately misread it for funnies
 
that's better than some real sites I've seen
it's legible, and I can find the info I want fast
except maybe their phone number
 
Jon
Date: March 31th, 2004
 
my mac is randomly closing software that's runni
 
Jon
it has become sentient
 
4:11 PM
@Jon the reflection in the spoon looks like they are taking the picture like this:
 
Jon
lmao
i did this for a class project in 2004
 
It totally does look like Yoga...
 
Pip
lol
hey guys..... may I have some help looks around for mods
 
Why mods?
@JohnMcDonald - good afternoon mate, Welcome back
 
Pip
@Blue jk, Noc and I had a convo yesterday about this
can I has some upvotes? I want to get to 500 rep so I can review
 
4:26 PM
begging is a surefire way to get a certain type of person to think twice before upvoting anything of yours
 
@Pip, If you want upvotes, make upvote worthy content. Although I appreciate people upvoting my content on here I only appreciate it if it's honestly good content. Otherwise it degrades the quality of the site.
 
tread carefully
 
Pip
@Almo alright
@Blue true
 
Blue makes a good point as well. :)
 
Create answers, link them, ask us to look. We will always look. It is up to us it is vote worthy.
 
4:28 PM
^
 
Pip
alright :D
(I need to learn more :P)
up to 390 rep, got another edit :D
need 110 more
 
#dyor
stands for Do Your Own Research.
my contribution to the post you edited.
 
Pip
alright
thanks
someone upvoted my famous question:
17
Q: Random/Procedural vs. Previously Made Level Generation

PipI am making a game (called "Glory") that is a top-down explorer game, and am wondering what the advantages/disadvantages of using random/procedural generation vs. pre-made levels are. There seems to be few that i can think of, other than the fact that items may be a problem to distribute in ran...

(most famous)
so old....
I sorta want to edit it
edited :D
(I guess I bumped it, too.... hmm)
 
81
A: Has piracy ever resulted in a developer getting shut down?

BlueAfter a little running around the internet I found an interesting article that shows a good example of how pirating can affect a game directly and shutdown a project. iOS Game, Battle Dungeon, Forced To Shut Down Due To Piracy In this article, Hunted Cow, the developers behind the iOS game Batt...

<3
 
Pip
nice
I remember that question
that was an answer, tho
 
4:43 PM
You get rep from answers too man
ALL THE THINGS
 
THIS CHANGES EVERYTHING
 
Pip
@Jon do you eat bacon THAt often while coding?
@ToddersLegrande ?
 
Sarcasm
 
Pip
ah
 
Jon
Yes.
 
4:46 PM
I love you @ToddersLegrande. I get you, maaan.
 
<3
I enjoy shouting in chat for some reason
 
You just love me that much you have telegraph your presence...
 
Pip
up to 397 rep :D
 
How many points per edit?
 
hi halp
someone closed my question, is it possible to answer on this site or is the question impossible to get to fit?
 
4:49 PM
@ToddersLegrande 2 I think
 
0
Q: Card Game AI Architecture

HarryWhat type of AI I should implement for a card game like "Magic: The Gathering? There are so many cards in "Magic", more than 3000, that have crazy, differing rules with no real consistency. I think writing a BT or any more traditional system would be too much work. It would already be a lot of e...

 
@JohnMcDonald So I only need to edit like... 750 or so questions to get close vote power!!! :O
 
Pip
@JoshPetrie I was about to flag that question :P
 
Sounds about right. heh, :(
 
Pip
@JohnMcDonald yep
@Harry I edited that, did it stay pretty much the same in meaning?
It will be reviewed, since I edited it
 
4:52 PM
@Harry I'm by far an expert on AI but I see a naive / simple approach and a far more complex approach
To be naive about it, we can look at our cards in hand and make best decisions based on that
To be complex, the AI should have knowledge of everything thats in the deck, what's he's played so far, and what's remaining so that it knows the odds of cards that synergize with his hand to be drawn
 
Pip
3
Q: How to encourage on hold questions to be improved by the asker?

concept3dThis is an issue that is bugging me, a lot of the on hold questions can be improved, yet I feel like most people don't get the idea that their questions will be reopened once they are improved. I think it's our fault we are not conveying this idea correctly. And they feel like it won't be opened,...

 
This latter option requires a significantly larger amount of data up front
 
Thanks for your edit Pip I think it made the question better.
 
Pip
np
 
And you can even take it a step further by giving the AI some knowledge, or at least the ability to guess about his opponents deck
 
4:54 PM
Would a NN fit the domain at all?
 
Don't know what that is
 
neural network
 
That's beyond my knowledge. Like I said... I'm definitely not an expert on AI or how its traditionally implemented
Just spouting off an approach I would take
 
okay thanks
I just wish the question could be opened... I don't think it's too opinion based or whatever.
 
Pip
@Harry It is being reviewed, it may or may not be opened (since I edited it)
 
4:56 PM
it's opinion based because there's no right or wrong answer
 
Pip
you could ask "list some possible AIs that I could use and their implementations"
 
List questions are off topic as well :P
 
Pip
:P
don't listen to me, then
 
You edited it again? I didn't get a notification.
 
Pip
I didn't
 
4:57 PM
Yeah the edit's already gone through.
 
Pip
bbiab
 
@Almo the right answer's the one that gives the most information to me I think?
 
user4704
No.
 
user4704
The right answer is the right answer. In this case, there isn't one. This site isn't just about you, it's also about creating a canonical database of questions and their answers for future users.
4
 
user4704
There are multiple equally-valid answers to that question (I would have closed it as "too broad," but opinion-based is reasonable enough).
 
user4704
5:00 PM
If you narrow the scope of the question to a specific problem you are having implementing one of the available options for AI, that would be on-topic.
 
Josh I've been using SE for years, there are so many "what should I do" questions on all sites.
 
user4704
Some sites have different tolerances for them than others.
 
I actually like the AI question and would be interested in reading up on an answer myself... but there are probably far too many approaches to be appropriate for the site :)
 
Josh so you're saying gamedev only answers "How" questions but not "Why" or "What" questions?
 
user4704
No, there are appropriate "why" and "what" questions.
 
user4704
5:04 PM
We do not consider questions for which every answer is equally valid appropriate. It's in the FAQ.
 
user4704
Your question, as asked, does not narrow the subject enough.
 
Basically "What should I do" is allowed only if the answer is non-controversial?
Any advice on narrowing it?
 
user4704
"What should I do" is almost never allowed, unless it's "what should I do to fix this error or specific problem."
 
Pip
back
 
user4704
Well, since just about every possible option for implementing your AI is equally valid as an answer to that question, I'd just pick one and try to implement it, and then ask questions based on any problems you encounter.
 
user4704
5:07 PM
I will say, however, that a neural network is almost certainly not the right approach.
 
Pip
hey @Grey
and @Byte56 and @mmyers
 
user4704
"AI" in games is almost always scripting or decision-tree stuff.
 
1 min ago, by Josh Petrie
"What should I do" is almost never allowed, unless it's "what should I do to fix this error or specific problem."
 
Pip
lots of people at once
 
In which case your question could quite easily be "How do I fix this error or specific problem"
 
user4704
5:08 PM
Academic techniques like neural networks rarely find much foothold; they are cumbersome to build and control in a way that results in fun.
 
I'm not a person, I'm an avatar that has gained consciousness. I'm here to ask you all to stop playing games where the avatar dies, you're murdering my fellow avatars.
4
 
user4704
@Byte56 lol
 
Pffft @Byte56 :D
 
@JoshPetrie What kind of what questions are allowed then? I mean, other than the rephrased "how" question haha.
 
user4704
Look around at (most of) the unclosed questions.
 
5:10 PM
Check out the FAQ and help section too.
 
Thanks for the comment on NN, I'm pretty sure that view is very 2012 though.
 
user4704
Yeah, no.
 
They have plenty of information on it.
very 2012?
As in resulting in the end of the world?
 
Like this question:
17
Q: Random/Procedural vs. Previously Made Level Generation

PipWhat the advantages/disadvantages of using random/procedural generation vs. pre-made levels? There seems to be few that i can think of, other than the fact that items may be a problem to distribute in randomly generated terrain, and that the generated terrain may look weird. The downside to pr...

is basically the same thing as what I asked only phrased a little differently
 
user4704
Nothing about the fundamental characteristics of neural network theory or its pros and cons has changed much in the last two years.
 
5:11 PM
What are the adv of NN vs BTs for a usecase like mine.
 
user4704
@Harry That question should probably be closed based on the title.
 
user4704
Remember that we are only human and cannot read and process every single question on the site.
 
@Harry Just so you know, pointing out other questions doesn't make yours more valid, it just argues the case that those questions should be closed too.
 
user4704
So eventually some do avoid closure that should otherwise be closed.
 
user4704
(you will also likely find old questions that should be closed since site policy was changed after they were asked, for example)
 
5:12 PM
@Byte56 I was genuinely unsure what the difference was rather than making an argument I htink.
 
@Harry
http://gamedev.stackexchange.com/help/on-topic
http://gamedev.stackexchange.com/help/dont-ask
That basically puts everything you've asked into context.
 
Jon
so how are the new guys suppose to make points?
 
@JoshPetrie I watched this video: gdcvault.com/play/1015667/Off-the-Beaten-Path-Non it seemed like a very successful implementation of NN
 
Jon
All the easy questions have been asked and answered.
 
@Harry I would vote to close the "Random/Procedural vs. Previously Made Level Generation" question, but I think it's borderline enough that it doesn't deserve a unilateral decision from a moderator.
 
5:15 PM
@Jon Ask about problems they are having and not what approach is best
 
@JoshPetrie in supreme commander
 
@Jon this site isn't about earning karma like Reddit, that doesn't matter.
 
user4704
@Harry I can't load that (no Flash), but the input domain there sounds much different.
 
Jon
It matters
 
You're not aiming for a control point so you can edit, moderate and review.
 
Jon
5:16 PM
my boss said I need 10K points before march, or i'm fired.
 
Then get a new boss
 
Jon
I'm my boss
 
@Jon 0.o
haha
 
Then you're doing it wrong.
The purpose of this site is to be better as a whole, not be a catalogue of who is better than who.
 
Yeah SE is like a wiki, rather than a community.
 
Jon
5:17 PM
if you get higher points, it removes ads and such though
And you get more power
 
@Blue more people with more points means more people with privileges which makes the community better as a whole
 
Wiki and community are not exclusive.
 
@ToddersLegrande not necessarily.
 
It doesn't care about "engagement" it just wants quality pages.
 
More people with close vote power means less unilateral close hammers from the mods and the community can do more to moderate itself
Same with all the queues that are available for review
Reopening questions, reviewing first time posts, etc.
 
5:19 PM
@JoshPetrie Is it that there's fewer inputs for supreme commander?
 
user4704
Probably, and a different class.
 
More people who care about the site = a better community as a whole. Rep allows opening up this moderation but in the same token we maybe should be looking at the milestones for moderation and close vote privillages
 
@Harry Check out Unity Answers if you want to see a site like SE that doesn't police their questions.
 
Can you go into more detail so I can understand it?
 
Full of junk, hard to find any good answers.
 
5:20 PM
@Byte56 god, I couldn't agree more.
 
user4704
Well, again, only speculating since I can't load the thing here, but some cursory research suggests that they use a relatively simplified neural network to base broad decisions on, but not specifics.
 
user4704
So they have inputs like the various states of their resources and outputs are things like priorities for certain classes of job.
 
user4704
I do not believe the inputs to the network are, for example, the position of all known enemy units and the outputs the specific orders. That isn't really how they work; training and scaling that would be basically impossible, it's too complex a decision surface.
 
user4704
By analogy you could use a NN in your card game to prioritize attack versus defense, maybe, but at that point it's such a simple decision that you can do it heuristically.
 
Well sure you have broad which feeds down to specific. Maybe at the very low levels you can script simple behaviors but high level decision making like when to attack when to run can be handled by the NN
Well no I can't, that's too hard.
The card game has too many rules / is too complicated for that to work.
The other possibility is a HTN I guess
Here's the pdf for the talk i linked you maybe you found it already twvideo01.ubm-us.net/o1/vault/gdc2012/slides/Summit_AI/…
 
5:26 PM
hi
can anyone help me to create algo for 3x3 puzzle ?
 
@Ketan deja vu?
Was it you that was asking about this before?
 
Pip
@ToddersLegrande I think so
 
@ToddersLegrande yes
 
@Pip You should stop doing that to every person that walks in here
 
5:28 PM
@Pip Truly, that's pretty rude.
 
Jon
yeah @Pip whats up with that
 
@Pip wth is wrong with you man.
 
I have thousands of replays that I could train my data on too.
 
I have done some logic but facing little problem
 
I think Pip got the message.
 
user4704
5:29 PM
@Harry Well, that approach is different than I thought but it still seems like one bit of a whole.
 
user4704
But like I said, you should try it and see the problems you run into.
 
user4704
My gut says that it's not going to result in particularly fun opponents.
 
It feels like a lot of investment.
 
user4704
But I could be wrong.
 
Is it just that it'll be too good?
 
5:30 PM
Trying and failing is one of the best ways to learn what you need to know.
 
Hello peoples
 
user4704
@Harry The results I have seen generally are "too good" or "too random," yes.
 
Pip
so...... many...... pings...... brain........ overwhelmed.....
 
I like watching others fail, and not doing that.
 
Pip
@ToddersLegrande alright
 
5:30 PM
Turn off the sound @Pip
 
^+1
 
Pip
I already did, a few days ago
 
@Byte56 I just wanted to get veteran feedback so I'm not doing really stupid things.
 
user4704
@Ketan If you have a specific problem that is on-topic for this site, you should ask it on the site itself.
 
Maybe your browser forgot and put it back on.
Home time for me boys, I'm off.
Enjoy your weekend if I don't get to chat with you.
 
5:31 PM
bye bye @Blue!
 
hugs
 
"Bye Bye Blues" is a popular and jazz standard written by Fred Hamm, Dave Bennett, Bert Lown, and Chauncey Gray and published in 1930. The year it was introduced it was sung by The Vikings on the NBC radio series, The Vikings. It has been recorded by many artists, but the best-known recording is one made in 1952 by Les Paul and Mary Ford. This recording was released by Capitol Records as catalog number 2316. It first reached the Billboard magazine Best Seller chart on December 27, 1952 and lasted 5 weeks on the chart, peaking at #5. Recorded versions *Oscar Alemán *Mose Allison (1958)...
 
@JoshPetrie I want to create algo to check move is valid or not in 3x3 puzzle.
@JoshPetrie I am using this code for it
 
user4704
"How can I determine a valid move in a 3x3 puzzle game" seems pretty on-topic.
 
Pip
yep
 
5:33 PM
@JoshPetrie I really don't think my question is that subjective. I found your comments really helpful and confirms a couple of things for me, I think a more authoritative "NO, NN is wrong for your domain", or "these would be the problems you would run into with NN so better use BT" would definitely be "correct".
@Byte56 I don't want a unmoderated site but some sites are imo way too strict.
 
user4704
In fact I'm pretty sure there is such a question already.
 
user4704
Oh, I guess I'm thinking of this: gamedev.stackexchange.com/questions/40035/… which isn't the same.
 
user4704
Nonetheless, @Ketan, it sounds like something you should try asking on the site.
 
user4704
Please explain what a "3x3 puzzle" game is as well for the benefit of readers who may not know.
 
Jon
tic tac toe
 
5:36 PM
@Harry The strictness here is based on the communities meta consensus.
We tried the less strict thing, it wasn't pretty.
 
Jon
Solve this tic tac toe grid
 
I have 3k points on stackoverflow it seems fine to me
 
Stack Overflow has sooo much more volume.
 
@JoshPetrie I am using libgdx so I have only positions
 
5:38 PM
Also a different meta community that has set up different guidelines on what to allow.
 
user4704
@Ketan Please ask the question on the site.
 
and I used this code but faceing problem when position=3 and emptypostion=4
if ((position+1)==emptyPos
||(position-1)==emptyPos
||position+3==emptyPos
||position-3==emptyPos)
 
user4704
... please ask the question on the site
 
Jon
lol
 
Also. It's a lot easier to amass high amounts of rep on SO, if you play the rep game right. Just as a factor of that volume. In the end though, the different SE sites are different.
 
Jon
5:39 PM
you may need to rethink your logical statement
 
if you'd like to change this site, please bring up what you think is an issue on meta.
 
@Ketan Ask the question on the site.
 
Pip
@Ketan ^
 
I think I might try to hire an AI consultant lol, would probably be easier than asking on the internet.
 
5:41 PM
@Harry Paying money is a better bet than free.
 
Jon
an AI consultant. now that is going to be enjoyable.
 
Pip
lol
 
Possibly, on hold also doesn't mean gone. It just means that there are issues with the question :\
 
Pip
:/
 
@Harry Hey Harry, I just got done taking an AI course this past semester and we actually did quite a bit of NN type of work. I left that class knowing that I really misinterpreted what a NN truly does when I went into the class. NN's Are tough to train, there are so many ways of training depending on what type of data you want to feed it, etc. Getting it to make good decisions on complex things such as moves in a game is very tough.
NN's are tough to get right. I would try implementing something tree-like for a card game and iterate on that. You will get somewhere faster that way. I tried creating an AI for checkers using a NN and in the end, I ended up with a decision tree doing the majority of the work anyways.
 
Pip
5:43 PM
 /\
. .
.__.
weird face
 
@DeanKnight thanks, checkers has way simpler heuristics than my card game though
One thing I should say about my game is that the rules of the game are user generated
 
Pip
I need 43 more rep :D
 
Would, "What difficulties would I run into using a NN for this" be a suitable question?
 
so it's actually impossible for me to predict what the cards should do
 
Pip
I have earned 77 rep on GD SE today <3
need more.....
 
5:45 PM
There's an entire wall of questions you could answer :p
 
Pip
where?
 
@Noctrine yes I think it would, I'm not sure if it would get me what I want to know but it would get me a lot closer probably.
 
Pip
.....
 
@DeanKnight What did you think they were and how was it different?
 
Pip
5:45 PM
I looked already :/
I will look again
 
You're ultimately trying to figure out if a NN is appropriate aren't you?
 
At all of them?
 
user4704
@Harry That makes it seem even worse for an NN.
 
Or are you dead set on doing a NN for it?
 
user4704
Since you don't have that data to feed to the training portion of the network?
 
5:46 PM
@Harry That seems like the absolute worst problem to have when designing machine learning based things :\
 
Jon
sounds like a bad case of overengineering
 
so much blue
 
I'm not but I need to clarify why I brought up NN and make sure people understand what the constraints are.
 
Pip
hey Mick
 
hi pip
 
5:47 PM
This is also a reason for the on hold case, sometimes a bit of discussion will help everyone in the end.
One blue down :p
 
Pip
lol
 
@JoshPetrie I have thousands, not far from million, of replays to feed into the training
of course they're all for widely different rulesets...
 
user4704
Yeah, I think that's going to result in an AI that appears to just do things randomly.
 
Pip
sounds fun
 
Jon
I can code that
 
user4704
5:48 PM
And how does that account for user-generated rules that were not in the set you have?
 
user4704
Or that were not sufficiently represented in that sample set?
 
I feel like I started getting all of the spam calls, when I added myself to the Do not call list :\
 
Pip
@Jon Python has a random module :D
 
to what length do people generate their own ruleset?
 
No the rulesets are part of the replays so each replay knows which ruleset goes with it etc, i can call up all the replays for a certain ruleset and have those run
 
Jon
5:49 PM
I have a real time dice shuffling machine, that spits out random numbers
 
user4704
@Harry Right, but you have thousands. Are all rulesets equally represented?
 
the rulesets are completely indeterminate, all I know is that there are such things as "cards"
 
user4704
What about new rules that are not in that set?
 
@JoshPetrie they're not, some will certainly have more training data than others
 
user4704
To what extent does a user generate new rules?
 
Pip
5:50 PM
should I have yogurt or cinnamon raisin toast?
 
Jon
if you don't have rules, you are creating chaos.
 
@Pip Yogurt
 
Pip
alright
brb
 
@Pip cinnamon raisin toast with peanutbutter and cinnamon sugar sprinkled on top
 
Theoretically, he could feed in new game replays as they are played. In order to keep it up to date over time. But changing the ruleset each time? Haha I dont know how you will create a proper NN with that going on (My knowledge in NN is only 1 semesters worth afterall)
 
5:51 PM
The rules are unpredictable, some games have hands and draws and putting things into play on the board, some don't, some have buildings which build cards, some have resources some don't, some have a oard some don't, some have grids which move things in 2d space some have only 1 area in which everything goes etc
 
@Harry do you work with NN much? Are you far along enough to break the task down into separate sections with high bias?
 
Do you happen to be in college/university? I would contact an AI professor and just chat about it. Honestly though, even if you figured something out, I would doubt its run-time efficiency. I think thats one of the biggest things holding complicated NN's back in games.. they can be pricey performance wise.
 
Pip
@ToddersLegrande hm.... I may have to do both :D
 
@MickLH I don't work with NN much but the users and game moderators can help with ruleset specific issues
 
Is this a client / server kind of thing?
 
5:54 PM
It's not about that, @Harry, it's a more technical aspect of the NN itself
Your classification archetypes must have high bias
 
It's a strategy game, think magic played over the internet.
 
Oh, well that makes it a lot easier.
It's not appropriate for what you want to do.
 
@MickLH can you explain what that means exactly?
 
Also the classification subject's bias is different from the usual bias hack
@Harry roughly, it means that there should be a "noticeable" difference between the patterns to classify
It's a pattern matcher not magic :P
 
Pip
@JoshPetrie nice edit
 
5:58 PM
@DeanKnight Not in an uni but if you could contact a AI prof that would actually be pretty amazing.
 
Hmm, well maybe as a research project. That'd be fun. A machine that could learn card games.
 
I am concerned about the runtime cost too
@MickLH I think there should be sufficient bias
 
I wouldn't worry about that too much, you can always throw a GPU cluster at it
 

« first day (1268 days earlier)      last day (3754 days later) »