Soup search results

[Hunting]

Really the lower bound would be much higher than that, because if one soup exists then there are probably gajillions of variants of it. Here's a 20x18 sample that can probably be packed into 16x16 with some effort:

Code: Select all

x = 20, y = 18, rule = B3/S23
8bo$8b4o$8bo3bo2$10b3o7$15b2o$bo13b2o$obo5bo$6b2ob2o$3o2bo3b2o7b2o$6b
2ob2o7b2o$6b4o!

19x17 parent:

Code: Select all

x = 19, y = 17, rule = B3/S23
7b3o$10b3o$10b3o$11bo7$15b2o$15b2o$3o6bo$b2o3b2obo8bo$2o5bobo8bo$5b3ob
o6bobo$8b2o6bobo!

16x16 soup:

Code: Select all

x = 16, y = 16, rule = B3/S23
o2bob2o$o3b2ob2o2bob2o$bo2bo4bo3bobo$10b4o$2b2obo3bobobo$o4b4o3bob2o$
4b2o2b3o2bo$b2o2bo2bo6bo$4b2o2b2o3bobo$3o2b2o2b2o4bo$ob2ob2o2b2obo$o2b
obo4b2obo$3ob2ob2o4b3o$o2bob2o2bo3bobo$3o4bo3b2ob2o$b2obo2b4obo2bo!

Nice find! HOW DID YOU FIND IT?

[Freywa]

Nice find! HOW DID YOU FIND IT?

Logic Life Search, of course.

[dvgrn]

19x17 parent:

Code: Select all

x = 19, y = 17, rule = B3/S23
7b3o$10b3o$10b3o$11bo7$15b2o$15b2o$3o6bo$b2o3b2obo8bo$2o5bobo8bo$5b3ob
o6bobo$8b2o6bobo!

16x17 parent following a different line of approach:

Code: Select all

x = 16, y = 17, rule = B3/S23
6b2ob3o$10bo$5bo2b2o$5b2o2bo$6b2o$4bobo$5bo4$11b2o$o10b2o$o$o$b2o11b2o
$b3o10b2o$b2o!

This is without doing any directed lifesrc searches, just using the known alternate two-object switch engine seeds:

Code: Select all

x = 155, y = 179, rule = LifeHistory
31.A$30.A.A$30.2A3$34.B$27.A5.2B$26.A.A2B.3B7.2A$26.B2A6B7.2A$26.8B$
25.D7B$25.2DAD4B$25.DA.ADB$27.2AB$27.3D$50.2A74.3D$50.2A71.A3BDA$122.
A.A2BA.AD$47.2A74.2A2BD2AD$47.A.A73.5BDBD$47.A75.6B$32.2A90.4B$32.2A
89.5B$20.3D99.4B$17.A3BDA98.4B$16.A.A2BA.AD95.4B$17.2A2BD2AD94.4B$17.
5BDBD93.4B$17.6B94.4B$18.4B94.4B$17.5B93.4B$16.4B94.4B$15.4B94.4B$29.
2A81.4B$29.2A80.4B$110.4B$109.4B$108.4B$107.4B$106.4B$17.2A86.4B$17.
2A85.4B$103.4B$102.4B$101.3AB$101.2BA$.2A99.A$A.A$2.A5$123.3B$122.6B$
122.4B3D3.A$118.2B.6BD2B.ABA$117.10BD2BABAB$116.6B2A3BD2B2A2B$115.6BA
2BA3BDBDB$114.4B2.2BABA4B$113.4B4.2BA5B23.AB$112.4B5.6B24.BABA$111.4B
9.B25.2B2A$23.ABA84.4B35.4B$23.B2AB82.4B35.4B$24.A3B80.4B35.4B$25.4B
78.4B35.4B$26.4B76.4B35.4B$27.4B74.4B35.4B$28.4B72.4B35.4B$29.4B70.4B
35.4B$30.4B68.4B35.4B$31.4B66.A3B35.4B$32.4B64.B2AB35.4B$33.4B.2B60.A
BA35.4B$34.6B97.4B$34.8B94.4B$34.8B4.B88.4B$35.8B2.2B87.4B$35.12B86.
4B$34.2B2A3BA5B85.4B$34.BABA2BABA3B84.5B$35.ABD2B2A4B83.5B$34.2AB4D5B
83.4B$36.BD3BD3B81.3D4B$36.9B80.B2A6B$18.2B16.3B3D4B79.B.A2B2A2B$18.A
BA15.9B79.3AD2BABAB$18.B2AB15.6B59.BA20.A3BDBDAB$19.A3B15.4B59.ABAB
21.B3.2B$20.4B78.2A2B$21.4B78.4B$22.6B2.2B.A70.4B$23.8B3A10.2A59.4B$
24.6BA7B6.2A60.4B$25.5B2A3D4B68.4B$24.9BD6B68.4B$24.9BD2BD3B7.2A60.4B
$26.5B2AD2BD3B7.2A61.4B$26.4BABABDBD3B71.4B$28.3BA7B73.4B$29.9B75.4B$
29.9B76.4B$28.6B2.B78.4B$27.4B85.4B$117.4B$118.4B$119.4B$120.4B$121.
4B$122.4B$123.4B$124.4B$125.3B$121.2A2.B3D$121.A4B.2A$122.3A2BA2BD$
123.BAB.D3A$124.B3.D.A12$107.AB$107.B2A$107.2A2B$108.4B$109.4B$110.4B
$32.A78.4B$31.A.A78.4B$31.2A80.4B$114.4B$115.4B$35.B80.6B2.2B.A$28.A
5.2B81.8B3A$27.A.A2B.3B7.2A73.6BA7B$27.B2A6B7.2A74.5B2A3D4B$27.8B83.
9BD6B$26.D7B84.9BD2BD3B$26.2DAD4B86.5B2AD2BD3B$26.DA.ADB88.4BABABDBD
3B$28.2AB91.3BA7B$7.2B19.3D92.9B$7.ABA41.2A71.8B$7.B2AB40.2A73.2B2.B$
8.A3B$9.4B35.2A$10.4B34.A.A$11.6B2.2B.A25.A$12.8B3A10.2A$13.6BA7B6.2A
$14.5B2A3D4B$13.9BD6B$13.9BD2BD3B$15.5B2AD2BD3B$15.4BABABDBD3B$17.3BA
7B$18.9B$18.9B$17.6B2.B$16.4B$30.2A$30.2A6$18.2A$18.2A!

[Macbi]

Nice find! HOW DID YOU FIND IT?

Logic Life Search, of course.

Correct.

[testitemqlstudop]

16x16 soup:

Code: Select all

x = 16, y = 16, rule = B3/S23
o2bob2o$o3b2ob2o2bob2o$bo2bo4bo3bobo$10b4o$2b2obo3bobobo$o4b4o3bob2o$
4b2o2b3o2bo$b2o2bo2bo6bo$4b2o2b2o3bobo$3o2b2o2b2o4bo$ob2ob2o2b2obo$o2b
obo4b2obo$3ob2ob2o4b3o$o2bob2o2bo3bobo$3o4bo3b2ob2o$b2obo2b4obo2bo!

next april fools troll wen

[Hunting]

16x16 soup:

Code: Select all

x = 16, y = 16, rule = B3/S23
o2bob2o$o3b2ob2o2bob2o$bo2bo4bo3bobo$10b4o$2b2obo3bobobo$o4b4o3bob2o$
4b2o2b3o2bo$b2o2bo2bo6bo$4b2o2b2o3bobo$3o2b2o2b2o4bo$ob2ob2o2b2obo$o2b
obo4b2obo$3ob2ob2o4b3o$o2bob2o2bo3bobo$3o4bo3b2ob2o$b2obo2b4obo2bo!

next april fools troll wen

Well at least Macbi has still done something nontrivial here, namely finding a IDK-tick predecessor of a 2EC in a 16-by-16 box.

[dvgrn]

16x16 soup...

Okay, so can someone check my math? I think this means that the absolute lower bound on 2EC probability is one chance in 2^256/8 = 10^76, more or less.

It seems reasonable take out a dozen or more orders of magnitude from that -- if a working soup is this easy to find, there should easily be trillions of variants that would also produce a 2EC. So 10^64 is still fairly conservative.

The first witch's first two lines in Macbeth add up 64 characters -- not counting all the scene-setting:

ACT I
SCENE I. A desert place.
Thunder and lightning. Enter three Witches
First Witch
When shall we three meet again
In thunder, lightning, or in rain?

Any properly equipped Shakespeare-typing monkey is going to be working on a typewriter with at least 30 keys, so typing just the first dozen words correctly would happen one time in 30^64 = 10^94-plus, or somewhere north of 10^111 if said monkey is supposed to produce proper capitalization as well as punctuation.

So we'd need an awful lot of monkeys to get into reasonable balance with 2EC odds. Recruiting all the bacteria on earth (in the 10^30 range) to take over the monkey business and simultaneously type ALL-CAPS SHAKESPEARE ... won't quite do it, even if we're satisfied with just those first two lines.

On the other hand, we're only creeping up on having apgsearched 100 trillion soups so far, 10^14, in C1 and G1 combined, so we'd have to have a separate computer for every individual atom in the Earth running Catagolue for a couple of years (at current speeds), before Macbi's specific soup or one of its trillion cousins would be particularly likely to show up.

Alternate Calculation
Hunting's line of investigation seems like a good one: given that we see pairs of switch engines show up in soups already, how long before the right pair shows up? Here it might be better to try to calculate an upper bound to go with the lower bound above. Looks like we've seen something like 100 distinct pairs of switch engines show up and survive in asymmetric 16x16 soups. (There are more like 360 soups producing something weird that Catagolue reports as a non-GPSE non-BLSE, but a lot of them are duplicates arising from small seeds, and quite a few of them are actually singleton GPSE or BLSE soups with an intermediate stage that consists of a single active glider and a bunch of stable junk.)

That's about one-in-a-trillion odds of getting any old pair of switch engines, but that includes switch engines that appear several hundred ticks apart, and/or with long distances separating them. We need one particular combination with exactly the right timing, out of say 10,000 possible relative locations and 1000 ticks' worth of relative timings.

That puts 2EC soup odds at somewhere around one in 10^16, ten million trillion. That's definitely a lot better than one-in-10^64 odds, but still it looks like we're allowed to be really, really surprised (and maybe even a little suspicious) if a 2EC soup shows up in Catagolue tomorrow. The above doesn't even bother to factor in the odds of the right junk showing up behind the switch engines, so it's a painfully optimistic upper bound.

... Or if anyone can point to where I've bent the rules of ballpark estimation all out of shape, please feel free to correct me.

[hkoenig]

The problem with the comparison to "monkeys on typewriters" is that in a language, the probability of the next character is dependent on the previous characters. Individual characters are not independent events. For example, in English, "bv" less common than "ng", "qxr" is much less common than "the", etc.

Much more interesting than random gibberish is to feed in to a process a large amount of text, analyze the combinations, then generate new text based on that stats just generated. Your result will resemble the source material. (Try entering chemical papers, for example, many of those resemble gibberish to start.)

I wrote a simple FORTRAN program to do that sort of thing in the mid-70s, after reading an article claiming it couldn't be done because there wasn't enough data storage to contain the necessary tables. The author assumed multi-dimensional arrays (27x27x...), while I implemented the data structure as a 27-way tree. Easy to go to about 8 letters deep that way, and avoid wasting lots of space on combinations that never happen. The hardest part was actually having to type in all my source text...

[dvgrn]

The problem with the comparison to "monkeys on typewriters" is that in a language, the probability of the next character is dependent on the previous characters.

I think I ran into this first in the book version of a Computer Recreations article. June 1989 was the original Dewdney article on Markov chains. Maybe Hofstadter mentioned it in one of his Metamagical Themas columns, too, like the ones about nonsense, I don't remember for sure.

Now we have neural networks doing the next generation of the same thing -- have you seen the relevant Sandbox thread? If not, my apologies for inflicting it on a wider audience.

I suppose Markov-chain monkeys would be a good modernization of the monkeys-on-typewriters meme. We could give the monkeys virtual keyboards where the keys change size according to the last several letters typed. Random keyboard mashing would suddenly be a lot more likely to produce snippets of Shakespeare ... but full scenes, even short ones, are still going to be a lot farther out of reach than a 2-engine Cordership.

[Ian07]

Natural cis-Coe's p8 on table:

Code: Select all

x = 16, y = 16, rule = B3/S23
bobooboobobbooob$
oobbboboobbobbbo$
ooobobbbobbbbbbo$
ooboobboooboooob$
obobboobbbobobbb$
bbobbbobbbooboob$
bbboooooooobboob$
bbobooooboooobbo$
ooooobbboooooobo$
obboooboooobobbb$
bbboooobooobobbb$
obbooobbobooboob$
ooboboobbobooboo$
obobbbbbobooooob$
ooobbooobbboobbo$
boboobobbobooboo!

Haul: https://catagolue.appspot.com/haul/b3s2 ... 819004caf2 (Rob Liston, 2020-06-08)

[calcyman]

Rob Liston found an infinite-growth soup which takes 10 514 926 generations to settle down into regular growth:

Code: Select all

x = 16, y = 16, rule = B3/S23
2o3bob4o2b2o$o3b2o5bo$o3bo2bo4b2obo$bo4bob2o2bobo$3b2ob2o3b4o$o3bo3bob
2ob2o$2b2ob2obob4obo$b10ob2o$2bo3b2o4bob2o$2o3b2o2b3ob2o$4obobo6b2o$ob
3o5bo2b2o$2bobobobo2b2ob2o$2bob2ob2o6bo$2b3obob5o2bo$bo3b2obo2bo3bo!

[dvgrn]

Rob Liston found an infinite-growth soup which takes 10 514 926 generations to settle down into regular growth...

Awesome! I've been hoping something like this would show up in an asymmetrical soup.

It's easy to synthesize, too. At worst, at T=406 it's basically just an R-pentomino plus some junk:

Code: Select all

x = 51, y = 58, rule = B3/S23
35b2o$26b2o7bobo$26bobo7b2o$27bo10b2o$31b2o5bobo$31bobo5b2o$32b2o$34b
2o$34bobo$35b2o3$24b2o$24b2o2$28b2o$28b2o$2o$2o4$8b3o$35b2o$34bo2bo$
34bo2bo$35b2o12bo$48bobo$49bo3$11b2o23b2o$11b2o3b2o17bo2bo$16b2o18b2o
15$36b2o$35b2o$36bo7$15b3o!

With slightly different junk some distance off in the southwest the stabilization time could end up millions of ticks longer or shorter.

[gameoflifemaniac]

Rob Liston found an infinite-growth soup which takes 10 514 926 generations to settle down into regular growth:

Code: Select all

x = 16, y = 16, rule = B3/S23
2o3bob4o2b2o$o3b2o5bo$o3bo2bo4b2obo$bo4bob2o2bobo$3b2ob2o3b4o$o3bo3bob
2ob2o$2b2ob2obob4obo$b10ob2o$2bo3b2o4bob2o$2o3b2o2b3ob2o$4obobo6b2o$ob
3o5bo2b2o$2bobobobo2b2ob2o$2bob2ob2o6bo$2b3obob5o2bo$bo3b2obo2bo3bo!

Wow, natural crystals! Also, the puffer is p5376, if my calculations were correct.

[calcyman]

Second natural occurrence of Rob's p16. This time it forms at generation ~ 2000:

Code: Select all

x = 16, y = 16, rule = B3/S23
o2bob5o4bo$4o2b2ob4o$bo5bob2o2b3o$o7bobo4bo$3b4obo4bo$o2b3o3b7o$3obobo
bo2b2o$b4o2b2obo$2obobob2o2bo3bo$b2obob3ob6o$ob3o5b2o2b2o$obobobo2b4ob
o$2b2ob3ob2ob3o$3b3o5b4o$o3b2ob3obobo$2ob6obob2obo!

[praosylen]

Second natural occurrence of Rob's p16...

The correct soup:

Code: Select all

x = 16, y = 16, rule = B3/S23
bbobboboboobboob$
oboboobobbooooob$
obbboobbobobbbbo$
obobbbbbbbbbbbob$
obbobbboooboooob$
oooobbbobboboobb$
obbooobbooboboob$
bobobbbbbboooooo$
bbooooobooobobob$
bbbbbobboobbboob$
boboooobobbbbooo$
bbobobbbbooobbob$
oobbooobbboboboo$
boobobbobbooboob$
oboooooobobbbobo$
boobobobbooboboo!

This should no doubt yield a different synthesis, but a reduction is a different question entirely.

[Goldtiger997]

Second natural occurrence of Rob's p16...

Rob's p16 in 15 gliders based off that soup:

Code: Select all

x = 83, y = 85, rule = B3/S23
81bo$80bo$bobo76b3o$2b2o$2bo7$44bo$43bo$43b3o18$48b2o$48bobo$49bo4$28b
o$29bo$27b3o7$41b3o14bo$25b2o16bo13b2o$24bobo15bo14bobo$26bo$11b2o21b
2o$10bobo20bobo18b2o$12bo22bo17b2o$55bo11$11b2o$10bobo$12bo2$74bo$73b
2o$73bobo12$b2o$obo$2bo!

[Ian07]

Final population 6768, found by Rob Liston on 2020-07-20 (haul unavailable):

Code: Select all

x = 16, y = 16, rule = B3/S23
ooooobbbbooboobo$
oooobbbbbbbobbob$
bobboobbooooobbo$
bobbooobbobbooob$
obooobbbooboobob$
obboobbboboobobb$
bboooboboobbbbob$
boobbbobobbbobbo$
oobbbbobbobbobbb$
ooobbooobbobobbb$
obbbobbooooobbob$
bobobbboboobbboo$
bbbooobbbbobbbbb$
boboooboobooboob$
bbobooobbbbobbbo$
bbooobbbobbbobbo!

[dvgrn]

Final population 6768, found by Rob Liston on 2020-07-20 (haul unavailable)...[/code]

Looking at things like these, I'm still surprised that Catagolue hasn't turned up a self-sustaining active junk-making reaction yet -- basically a new kind of switch engine, except maybe orthogonal or even oblique, and no doubt higher period.

I hereby predict that one of those will show up at some point, and everyone will be very surprised.

[calcyman]

Final population 6768, found by Rob Liston on 2020-07-20 (haul unavailable)...[/code]

Looking at things like these, I'm still surprised that Catagolue hasn't turned up a self-sustaining active junk-making reaction yet -- basically a new kind of switch engine, except maybe orthogonal or even oblique, and no doubt higher period.

I hereby predict that one of those will show up at some point, and everyone will be very surprised.

There are three of these in b38s23 (one orthogonal, one diagonal, and one oblique).

[goldenratio]

Fifth natural nonstandard non-flotilla spaceship - a MWSS and HWSS pulling a block:
https://catagolue.appspot.com/object/xq ... crf7/b3s23

Code: Select all

x = 16, y = 16, rule = B3/S23
ooboboobobbooobb$
boobbboooobbbbbb$
bbobooobbbbooobo$
obboooobbbbobbob$
oboobobboooobbob$
bbbbbbobbobobbbb$
oobobobbobboobbo$
bbbbboooobbobobb$
bbboobobbbbobobo$
bbbboobbobobbbob$
oobobbbbobobbboo$
boooboobbbbobbbo$
oobbbbooobbobbbo$
oooooooboobbbbbo$
obboooboboooobbo$
bobobbbobbbooooo!

Found by rliston on 2020-08-05.

[toroidalet]

Rob Liston found another loafer in G1 3 days ago (it forms extremely quickly; the first phase of the ship appears in generation 7):

Code: Select all

x = 16, y = 16, rule = B3/S23
bbooooobbobbbboo$
bboboooobobbboob$
bbbbbbbbboboooob$
obbbooobobbooboo$
bobboboboobobooo$
obbbbobooobboooo$
boobbbobbobboobb$
obbobooooboboboo$
booobbbbboobbbob$
obobobobooboooob$
oboboooooobbboob$
obboooboooboobob$
boboobbobbbooobo$
bobbbooobboobbbb$
bbbobbobboobobbb$
oboooobbbbbobbob!

[HartmutHolzwart]

Ther is a natural yl96 in G1:

http://catagolue.appspot.com/object/yl9 ... 5e09/b3s23

[dvgrn]

Ther is a natural yl96 in G1:

http://catagolue.appspot.com/object/yl9 ... 5e09/b3s23

Apparently it takes a very rare combination of switch engines to produce a yl96, since we don't even have one of those in C1 yet -- even though almost thirty times as many objects have been catagolued for that symmetry.

[calcyman]

Ther is a natural yl96 in G1:

http://catagolue.appspot.com/object/yl9 ... 5e09/b3s23

Apparently it takes a very rare combination of switch engines to produce a yl96, since we don't even have one of those in C1 yet -- even though almost thirty times as many objects have been catagolued for that symmetry.

The number of soups is the important metric: C1 has 38 trillion soups and G1 has 87 trillion soups, so G1 is over twice as large. That's why there are 778 cis-twin-bees-shuttles in C1 and 1689 of them in G1.

The reason that the reported object count is lower for G1 is that 'uninteresting' soups (those that quickly settle into p1/p2/p3/p6 ash with some combination of escaping standard spaceships) aren't censused, so each contribute zero objects to the census.

[Moosey]

Interesting!
I wonder when a 2EC will pop up, considering that it's also a (rare) combination of two SE's.