Slashdot Mirror
NES Games and Statistical Analysis
szadig writes "The New Gamer has published an article which tackles the topic of averaging gameplay. The accompanying video features 15 different players simultaneously shooting their way through the first stage in the NES game Gradius. From the article: 'The average time taken to kill the end level boss was 20.055 seconds, with the fastest player finishing him off in a mere 10.01 seconds. Six people finished the boss off at nearly identical moments. It would seem that the boss, bored with the player, actually self-destructs after 27 seconds. Beyond the almost perfectly synchronized explosions, further proof of this self-destruction can be found in the videos: no 10,000 point bonus (given to players when the boss is defeated) was awarded to these six players and, in a few of the runs, the boss detonated when there wasn't a single bullet near it.' Can we apply other statistical methods to gameplay?"
Getting 15 people together to do a statistical analysis to find out whether a boss self-desructs isn't quite as efficient as just having one person avoiding the boss for a while and just seeing whether it blows up or not.
A lot of the other stuff in the article was interesting though... looks like he put a lot of effort into it.
"When the atomic bomb goes off there's devastation...but when the atomic bong goes off there's celebraaaaation!"
I'd make a torrent, but I'm running Gentoo on an 800Mhz Athlon and it's gonna take me a couple of hours to compile Bit Torrent :-/
If the site dies.. I'll give my bandwidth a go:
http://www.wartsworld.com/AveragingGradius.mov
I would like to see an analysis of the original Sonic the Hedgehog's first stage. No matter how many times I play it, it seems exactly the same. I wonder how many variations there actually are!
Colonel Cranium this is Rectal Reconnaissance, we are on a collision course sir, Abort Abort!
So, that video made me wonder. Are there any multiplayer sidescrolling space shooters that allow up to 15 players to have a go at a level at once? That would be awesome!
It would seem to me that you'd need a sample size bigger than 15 in order to be considered "stastically significant". I would recommend conducting such tests with a far larger group of testers, or at least with the same people more than once so that you gather enough results to be somewhat conclusive.
2 cents,
Queen B
HDGary secures my bank
I've got 9.21 MB so far of 25.81 MB, only downloading at ~10 KBs :(
The full 26Meg file, in a decentralized azureus torrent:
t /announce
dht://1ACB2F69B008DAA48210AE53C3B96A8DE88C7B55.dh
Have fun!
Another interesting "average", though technically harder to pull off, would be to get 15 players simultaneously watching the same game in real-time, "averaging" in some reasonably manner the 15 inputs coming in, and feeding that to the game. It would be interesting to see if it sucks, or manages to play better than the individuals, or what.
I had never played a football game before so one day I played a friend of mine. Huge football fan and really into the electronic games.
Now this one was pretty standard fare you choose a play they run it, you throw to the player, now after watching for a while it became obvious that it was totally scripted, each receiver would be open at a set point in their run and you throw them the ball.
When I mentioned it was simply a matter of watching 2 receivers (maybe the blitz is too quick for the first one) and hitting the right button it made him really upset. He was totally commited to the idea that each play was unique with the game generating random behaviour.
People who play a lot of one game get a really good idea of how it's going to act given any stimulus, what this movie shows is how the game uses semi random behaviour in conjunction with scripted behaviour to give the illusion of thought.
On the one hand it's cool to show how to make a game seem to react like it's thinking... on the other, we haven't come far from pong.
This is very interesting. I played a game almost identical to this on the NES called "Life Force". Can anyone shed some light on why "Life Force" and "Gradius" look nearly identical?
Someone is just now discovering "the pattern?!"
Anyone who played these games KNOWS its not a matter of "figuring out how to outsmart or out-maneuver the end-level boss..."
It's always been just a matter of memorizing the pattern. Like, no shit.
This space available.
Why not disassemble the game and take a look at what the program is designed to do?
Dan East
Better known as 318230.
I think having 15 people on the same screen might create confusion, unless all the other players are faded and in the same color, or something like that.
The movie is very pretty, and its a step in the right direction. This sort of thing should be done much more.
However, 15 is not significantly significant. In fact, having an average time or counting who took the pickup doesn't really say much. In fact, the whole article doesn't say much because it needs a hypothesis.
I remember some retail outlets mapping the path people took in their stores in order to optimise product placement. That sort of thing could be done easily to create intuitive environments.
Or maybe its already being done.
$0.02
If the dependencies aren't too heavy, Azureus can be gotten as a binary. Unfortunately, my Gentoo box isn't booted right now, so I don't have the exact name, but you should be able to find the package without much trouble. Otherwise, thanks for sharing your bandwidth with Slashdot.
I've noticed for years that most console games follow a predictable pattern. The same enemies are always going to be in the same place every time a player runs through a level. These games perform predetermined actions actions at set times. PC games have generally offered more randomization and AI. I assume it's due to a lack processing power in consoles.
In recent years that has changed, but console developers still seem to have the habit of resorting to predictable patterns. While it adds a puzzle-type element to games it does make the game feel less dynamic. It probably requires less effort on the part of the developer and likely is easier to ensure good gameplay balance.
Anyway, seeing these patterns in those old games, while somewhat interesting, is no surprise. It also means that players are likely to handle encounters in generally the same way as there is really little effective alternative. There is generally an single optimal way through a game. Deviate from that path and you'll be met with varying levels of success, or failure.
If you want another example of statistical analysis in games, see this paper:
t ation.pdf
Is Bayesian Imitation Learning the Route to Believable Gamebots?
http://www.cs.yorku.ca/LAAV/pubs/file_BayesianImi
Abstract:
As it strives to imitate observably successful actions, imi-
tation learning allows for a quick acquisition of proven be-
haviors. Recent work from psychology and robotics sug-
gests that Bayesian probability theory provides a mathemat-
ical framework for imitation learning. In this paper, we in-
vestigate the use of Bayesian imitation learning in realizing
more life-like computer game characters. Following our gen-
eral strategy of analyzing the network traffic of multi-player
online games, we will present experiments in automatic im-
itation of behaviors contained in human generated data. Our
results show that the Bayesian framework indeed leads to
game agent behavior that appears very much human-like.
Did submitter not read the footnote for '**'? That player didn't win in 10.01 seconds, they died halfway through the level. Didn't you find it odd that the fastest time also had the lowest score? I can understand people replying who don't do it, but I would think that at least the submitter could take the time to RTFA.
Unpleasantries.