Talk:List of long-lived methuselahs

From LifeWiki
Jump to navigation Jump to search

Inclusion of methuselahs on this list

I propose that methuselahs only be included on this list if 1) they are 20 cells or smaller and 2) they last at least 1000 generations. If people want to discuss these criteria or some new criteria, please feel free to do so. Nathaniel 23:43, 9 February 2009 (UTC)

To be honest, the name of this list makes me think that it should only include the single longest-lived methuselah of each cell count. Then just the 20 cells criterion would be enough with no need for generational discrimination. That or just rename it "List of long-lived mesthuselahs" because that's closer to what it is right now, I think. Elithrion 00:17, 10 February 2009 (UTC)
Ah, good call. I think I'll go the "list of long-lived methuselahs" route then. Nathaniel 01:38, 10 February 2009 (UTC)
But still nothing seems to have happened in the area of cleaning out the list to meet any such criteria. We certainly need some criteria, otherwise we would have such things as:
  • somebody discovering an arbitrary large (in bounding box and/or initial population) pattern to beat what's here already (as seems to have happened already with Fred et al)
  • fun with a glider flying towards a block, which could start arbitrarily far from it and make the lifespan arbitrarily high while keeping the initial population down
Smjg 16:03, 12 June 2011 (CDT)
What is the purpose of this page? If it's to list any methuselahs that last a long time (however that is defined), then there are a whole heap that I can contribute that should be included here. If, on the other hand, it is to keep a record of methuselahs that last the longest time for a given cell count, then I propose the following be removed as they are all superseded by others on the list:
  • Ten
  • Syringe
  • Wake
  • Switch Engine
  • Multum in parvo
  • Rabbits
  • Bunnies
  • Blom
  • Justyna
  • Iwona
  • Eve
  • Ed
  • Edna
  • 35161M
  • 35201M
  • Fred
What do others think? Edwin 08:11, 24 November 2014 (UTC)
I'd say this is one of several areas on the LifeWiki -- GoEs and Corderships are two other cases that come to mind -- where we seem to have collectively decided to track named patterns that held some kind of record at some point. Once a record-holding pattern exists, it doesn't get demoted to un-notable just because it gets beaten by a new record-holder. Basically those are the patterns that have gotten talked about most, and therefore that someone might want to look up on the wiki. Also keeping track of them all provides some level of documentation of the historical progression of discoveries in the field.
That would mean that new record-breaking patterns should certainly be added to this list, whereas non-record-breaking patterns probably shouldn't be added. But patterns already on the list should not be removed, even if they're not current record-holders. (Best of both worlds -- not a lot of changes to make!) Dvgrn (talk) 11:54, 17 August 2017 (UTC)

About 30000 generations?

I posted this one 19 months ago in comp.theory.sci-automata:

#N Ewa (Eve)
x = 16, y = 10

I think it runs for some 30044 generations before stabilizing. 15:07, 14 February 2010 (UTC)

Wow, that's pretty impressive (by the way, it actually runs for 30046 generations); I'm sorry that it's been missed by the general Life community. How exactly was it found? do you still have the original pattern from which it came? (I'm assuming that it was found by testing large numbers of random patterns) Here is a link to the post alluded to above.
~Sokwe 19:57, 14 February 2010 (UTC)
I'm not sure now which option of my search program was used, but it was almost certainly generated through testing random walk patterns; when a promising one was found, it was also tested in combination with another predefined pattern (in this case, a hexomino) in various relative positions.
The random walk patterns were created by taking a number of random steps (8 directions allowed) and setting the cells on with a given probability (around 0.8). 14:31, 15 February 2010 (UTC)
I just added a page for it, and it to the list. Surprised nobody else has. Keiji 16:36, 11 May 2010 (UTC)


Maybe it would be a good idea to add ratio statistics to this table. I did the first three rows, then figured I should probably put it here rather than do all of them in case it's found to be overkill and deleted (or the like). I'd intended to make a table like this (as a sidenote, I was originally going to do L/F instead of F/L, then realized that F could, potentially, be zero):

Lifespan (L) Pattern Initial population (I) Final population (F) L/I F/I F/L
31082 Edna 26 3602 1195.4 138.5 0.038
29055 Lidka 13 1623 2235.0 124.8 0.056
28786 Iwona 19 3091 1515.1 162.7 0.107

If the Wikipedia javascript for sorting tables could be added this would be very useful too - IMO the L/I ratio is far more important than lifespan alone. Keiji 15:27, 11 May 2010 (UTC)

It could do with a column for bounding box, as well. Keiji 16:36, 11 May 2010 (UTC)
Does the table need columns for discoverer and/or year discovered? Lewis 17:55, 17 May 2010 (UTC)
I added a column for the bounding boxes. Discoverer/year discovered is probably unnecessary clutter here, I think, but the ratios could be interesting. (BTW, sortable tables are easy, just specify class="sortable" in the table header.) -- Schneelocke 21:32, 17 May 2010 (UTC)
Let's see, the L/F of diehard is 9999999999999999...(Infinity 9's later)...999999999999999999. -wwei23 9:56PM 10/20/2015 NY time

35000+ generation methuselahs=

Would any of the the methuselahs in this forum topic ( qualify for this list? Lewis 17:55, 11 May 2010 (UTC)

I think I found one

I was playing with The Game of Life when I found this methuselah that takes 1102 generations to stabilize.

This is generation 2 of the R-pentomino, a quite well-kno'n case. --Tropylium 16:23, 19 June 2011 (CDT)
Generation 1 of R-pentomino, not generation 2 of R-pentomino.

0100 0110 1001 0110

-wwei23 9:58PM 10/20/2015 NY time

40514-generation methuselah not on list

Here ( is a methuselah with lifespan 40,154 generations found by knightlife on the forums. Does it meet the criteria to be added to the list?


The RLE:

x = 78, y = 54, rule = B3/S23 75b3o21$o$o$o20$b3o6$9bo9b2o$9bo10b2o$9bo9bobo2$20bo!

This is a parent of 40514M. It runs for 40515 generations. Does it count? -wwei23 6:39PM 10/21/2015 NY time

Would this count?

. . * * * . .
. . * . * . .
. . * * * . .
. . . * . . .
. * * . * * .
. . . * . . .
. . . * . . .
. . * . * . .
. * . . . * .
* . . . . . *

Would that fit on the list? As you can see, it fits in a 7x10 bounding box and has an initial population of 21 cells. It lasts 2382 generations, and its final population is 448 cells. This isn't a joke (I can see what this looks like; someone joking around about a stick figure). I really was bored and decided to put some stick figures in and this one just happened to go for over 2000 generations. Feel free to test it out yourself.

The lifespan of notable methuselahs is linked to pattern size. Less than 2500 generations from 21 cells is not particularly remarkable, considering how already Acorn, with 7 cells, runs over twice a long. The amount of 8-cell patterns exceeding 2500G is in at least triple digits… so a very careful approximation for 21-cell patterns capable of the same would be 10²⁸. --Tropylium 15:52, 26 April 2012 (CDT)
An observation: it is easy to subtract a couple of cells from this pattern and end up with a methuselah that is about as good in terms of lifespan, or indeed even better. This seems like a sufficient heuristic for calling a methuselah non-notable.
x = 7, y = 10, rule = B3/S23
↑ 19 cells, 2518 gens, final pop. 364/369 (outputs a HWSS!)
x = 5, y = 9, rule = B3/S23
↑ 18 cells, 2060 gens, final pop. 298
x = 5, y = 9, rule = B3/S23
↑ 16 cells, 2038 gens, final pop. 320
x = 5, y = 9, rule = B3/S23
↑ 16 cells, 1879 gens, final pop. 356
I'm not going to make guesses on how many of the 15-cell patterns that could be shaved down from this pattern would also exceed 2000 generations, but probably there would be quite a few of them. --Tropylium (talk) 14:59, 27 November 2014 (UTC)

Can I add this? What should I call it?

A while ago I found this nice and interesting methuselah based on a collision between a glider and a HWSS. I modified it so it couldn't have been put at a farther distance, and it fits in a 8x9 box, and also is only 15 cells. It lasts for 11442 gens.

x = 8, y = 9, rule = B3/S23


It doesn't have a name though, so what should it be called? R-ship maybe?

The HWSS/glider collision has been known for long and it also gets periodically rediscovered. I think it's notable for being the longest-lived "two spaceship constructible" methuselah. The edited HWSS/R collection though, maybe not so much (there are probably longer-lived similar ones).
This argument would suggest that the glider-blinker mess and 2-glider mess (with relatively common names), which we currently lack, are also notable. By this naming scheme we could call the HWSS/g pattern "HWSS-glider mess". --Tropylium (talk) 15:11, 13 September 2017 (UTC)

also this:

x = 11, y = 11, rule = B3/S23


This one lasts til 10267, so slightly less, and is 16 cells, but this one I have no idea what to call it. --AwesomeFace 03:41, 25 March 2014 (UTC)

Double trouble

Two R-pentominos. Lifetime is 2501 generations.

x = 3, y = 13, rule = B3/S23

-wwei23 2:20PM 10/12/2015

Natural LWSS!

This one is Acorn+HWSS, lifetime=6643, and it emits a natural LWSS! Twenty cells.

x = 35, y = 18, rule = B3/S23

Slightly smaller version, 17 cells:

x = 35, y = 18, rule = B3/S23

-wwei23 10:01PM 10/20/2015 NY time

Addition of "historic" methuselahs

I've added a couple of "historic" methuselahs that we have articles on (basically, anything in Category:Methuselahs that wasn't already on the list). The rationale is as follows:

  1. we already have articles on these patterns, and consider them notable;
  2. they are of historic interest, especially compared to subsequently-discovered long-lived patterns; and
  3. the table collects and compares additional information not present in the individual articles, e.g. the various ratios.

I would like to point out, of course, that this isn't an invitation to add more small patterns with a lifespan of less than 1000 generations to the table. :) (Unless perhaps they're patterns identified in the early 1970s that the LifeWiki somehow missed until now.) Apple Bottom (talk) 21:51, 12 February 2018 (UTC)

Maybe a "year of discovery" column could be added to the table? That would make their significance clearer. 77topaz (talk) 21:54, 12 February 2018 (UTC)
Good idea, though I'm also thinking we should avoid duplicating too much infobox info, especially info that's not of a technical nature. Apple Bottom (talk) 22:06, 12 February 2018 (UTC)

I think I found one

It has 73 cells and lasts 19695 generations. It is derived from the following soup:

x = 128, y = 2, rule = B3/S23

But its cell count can be improved:

x = 126, y = 9, rule = Life

Also, if you think it isn't worthy, then remember that we have pages on at least two other soup methuselahs. -wwei23 11:48AM 3/18/2018 NY time

Not remarkable, no -- there's a 16 (!) by 2 methuselah on the list with a lifespan of 23662 generations already --, but don't take my word for it; post it on the forums and see what others think. Apple Bottom (talk) 20:33, 18 March 2018 (UTC)