Talk:2293: RIP John Conway

Explain xkcd: It's 'cause you're dumb.
Revision as of 04:24, 14 April 2020 by (talk)
Jump to: navigation, search

This is really very impressive. The design of the stick figure to allow it to release a glider that ascends upwards (the "soul" rising to "heaven" or whatever) with the body decaying - that's a hard thing to get right using just the Game of Life rules. 17:49, 13 April 2020 (UTC)

Although Randall is clever, the Game of Life has been studied for so long that I'm sure this is a well-known animation. Barmar (talk) 18:29, 13 April 2020 (UTC)
I played with the game a bunch in the past, but I've only done a bit of research after this appeared. I don't immediately find any previous report of this starting arrangement
This is unlikely, as the Game of Life has an uncountable number of patterns of this size, some of which are still being discovered. The pattern above is 7 cells wide by 9 cells tall - the number of distinct patterns that can be drawn in that box nears 2 --> 60 --> 1 (2^60). It's most likely that patterns such as this one are commonplace, and Randall just fiddled around until he reached one that he desired. The pattern itself, however, has likely never been discovered before. (As a fun postscript, a notable 8-cell wide, 9-cell tall pattern oscillating at period 16, just slightly larger than the one above, was discovered in February 2020.)Hdjensofjfnen (talk) 21:31, 13 April 2020 (UTC)
I suspect it's probably more likely that he worked it out backwards, rather than "just fiddled with it." 04:24, 14 April 2020 (UTC)

was there a placeholder comic posted before the gif went live?

Looks like this is second animated comic in xkcd, besides 1116(though 1190 could be possibly counted together) 19:34, 13 April 2020 (UTC)

Uh, 1331: Frequency and 1264: Slideshow immediately come to mind, and then I remember about 961: Eternal Flame. There's a lot more than two. Volleo6144 (talk) 19:43, 13 April 2020 (UTC)
Category:Comics with animation; just added it. Yngvadottir (talk) 20:19, 13 April 2020 (UTC)

The explanation says that the simulation is run on an infinite grid, but even when the grid is calculated out beyond the border of the viewable area, bounding errors & boundary formations can occur. I've never seen any implementation that actually produces an infinite, nor even practically infinite grid. (In fact, wasn't there a Minecraft mod that runs until it lags out the engine?) Can anybody point me to a truly infinite grid implementation? Conway's definitely was not infinite, he even commented at length about the boundary formations that show up at the grid edges (which are among the most subjectively beautiful, incidentally). I think the explanation needs correction? ProphetZarquon (talk) 21:59, 13 April 2020 (UTC)

The explanation is describing Conway's game of life, not any particular implementation IMHO. No change needed IMHOA. Also, the previous post was not signed properly. 21:47, 13 April 2020 (UTC)
Except many of Conway's observations about the "game" & even his initial description of it explicitly state & indeed hinge upon the fact that it's not (& thus far cannot be) implemented as an "infinite" grid. Part of the whole point of his experiment with it & the various demonstrations is to illustrate edge effects resulting from a finite range of calculations. It's extremely relevant that it's not infinite. It's actually kinda the whole point of his creating it, much the same way people working with fractals likewise tend to become very interested in bounding errors. The boundaries are where the interesting work is done. Apparently someone agreed with me at least in part, because they edited the wording. Thanks... Brian? I think we should actually add to the description of the "game" info highlighting the edge effects, because that's the primary focus of the project & its outgrowths in the first place. (We can't adequately simulate infinity & that's a big part of the interest in it.) Also, I frequently have to submit & then refresh & sign afterward because of the device I'm on. In this case I'm glad I did, because I saw your reply & the other new stuff!
ProphetZarquon (talk) 22:17, 13 April 2020 (UTC)
Golly uses a grid of arbitrary size by default. It can very easily be verified to at least ±210000. (Note that it also includes finite rectangular and toroidal grids.) LegionMammal978 (talk) 01:39, 14 April 2020 (UTC)

No computer [Citation Needed] can run an actual infinite grid, but with some intelligent bounding you can mitigate early signs of problems by maintaining "bubbles" of cells with offsets. You get into problems once you start machine-gunning out gliders (offsets will eventually overflow or awkwardly lose precision, depending on the var-types used; and maintaining a longer and longer bubble, or more and more bubbles just above glider-sized, is probably your other challenge) but it's probably good enough for most purposes. If you somehow have finite patterns that move out in huge (wasteless) cycles from the 'origin' and hold that path until enacting hugely-delayed doglegs (mathematically, it must be a point no further away than can be reasonably enumerated by the bits of information contained within each formation, and significantly less as it'd be a far less efficient count-down cycler than any folded LFSR, but it's imaginable) to meet again at some arbitrary (though deterministic and replicable) distance out in the far far reaches of your abstracted bubble-land then it's possible you could pretend you have infinite space. 22:26, 13 April 2020 (UTC)

ProphetZarquon, do you have a source for your claim that the main point of Conway's creation of Life was to study the edge effects? The page for Life on its own wiki describes the Life grid as "infinite" and only mentions edge effects as inaccuracies to be avoided, and this Conway interview about Life's creation doesn't mention edges at all. 23:34, 13 April 2020‎ (UTC)

First of all, as a tip, reply directly to the comment when you want to talk with someone. Secondly, yeah, I'm pretty sure that person's just blasting ass ham. 02:49, 14 April 2020 (UTC)

Possibly worth noting is the bit of artistry in the rendering. Munroe alters the step period of the iterations so that the deconstruction of the humanoid shape happens more quickly, with the stepping of the glider translating away occurring more slowly. Fixer (talk) 21:52, 13 April 2020 (UTC)

I like the comic very much, and I'm afraid to say I hadn't heard of his death in amongst... well, everything else. RIP. While I'm here, though, I'm a bit concerned about the current cell generation cycle explanation, as it feels awkward. Currently it is (paraphrased) "live cells survive if just enough neighbours / dead cells come to life with exactly enough neighbours / any other dies or stays dead". I'd prefer something that more delineates it as birth (dead to live, by propogation from the right number of live neighbours), death (live to dead due to either isolation or overcrowding) and continuation of state in all other cases. Can't work out a good phrasing yet, but may try it out later. 22:26, 13 April 2020 (UTC)

Agreed 100%. I believe the problem lies in the confusion between "cell as in biology" versus "cell as in jail". It would be better to avoid the word "cell" and describe a grid with squares that are either inhabited or empty; inhabitants with two or three neighbours survive to the next generation, otherwise they die (square becomes empty the next generation); exactly three neighbours to an empty square will give birth (become inhabited) in the next generation; and any other empty square stays empty.