### Re: Thread for basic questions

Posted:

**January 4th, 2019, 11:45 am**I assume you're excluding the rules b1357s1357, b1357s02468, b02468s1357 and b02468s02468?danny wrote:Can a universal GoE working in all outer totalistic rules exist?

Forums for Conway's Game of Life

https://conwaylife.com/forums/

Page **51** of **68**

Posted: **January 4th, 2019, 11:45 am**

I assume you're excluding the rules b1357s1357, b1357s02468, b02468s1357 and b02468s02468?danny wrote:Can a universal GoE working in all outer totalistic rules exist?

Posted: **January 4th, 2019, 12:11 pm**

Every GoE (a pattern that can't appear after generation 0) contains an orphan (a finite region of alive and dead cells such that any pattern containing that region is a GoE). It's possible to make a pattern that contains every finite arrangement of cells (just list all finite arrangements one after another). This pattern is therefore a GoE in every rule with a GoE.

However not all outer totalistic rules have GoEs. For example in B/S012345678 every pattern is its own parent.

However not all outer totalistic rules have GoEs. For example in B/S012345678 every pattern is its own parent.

Posted: **January 4th, 2019, 12:39 pm**

No, because those produce 4 copies of each pattern at the least.muzik wrote:danny wrote:I assume you're excluding the rules b1357s1357, b1357s02468, b02468s1357 and b02468s02468?

Good point about B/S012345678 though.

Posted: **January 4th, 2019, 4:27 pm**

Rules like B1357/S1357 and the other three *don't* have GoEs, because they are reversible; any finite region containing a single small pattern plus a large amount of empty space will have a parent that is very dense in that region, dying off in all places except for where the small pattern is (and what happens outside of the finite region is then irrelevant). If I recall correctly, there was a discussion about this a few months ago, where someone produced an example of this.

Posted: **January 4th, 2019, 4:28 pm**

Not sure why the number of copies would matter to the GoE-ness of a pattern. That just makes it really easy to prove that any given finite region of cells in those rules -- likedanny wrote:No, because those produce 4 copies of each pattern at the least.muzik wrote:danny wrote:I assume you're excluding the rules b1357s1357, b1357s02468, b02468s1357 and b02468s02468?

Code: Select all

```
x = 17, y = 7, rule = B1357/S02468
b3o3b3o2b5o$o3bobo3bobo$o5bo3bobo$ob2o2bo3bob4o$o3bobo3bobo$o3bobo3bob
o$b3o3b3o2b5o!
```

Code: Select all

```
x = 39, y = 40, rule = B1357/S02468
ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob
2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob
2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2o
b2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob
2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob
2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6o
b2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obo
bo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob
2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b
2o5b6ob2ob2ob2obo$b2ob2ob2ob2o2bo4bo4bo2bob2ob2ob2obo$b2ob2ob2ob2o7bo
4b2obob2ob2ob2obo$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$2ob2ob2ob2obob
3obo2b2o5b2ob2ob2ob2o$b2ob2ob2ob2o2bo4bo4bo2bob2ob2ob2obo$2ob2ob2ob2o
3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob
2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob
2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob
2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob
2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2o
b2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2o
b2o$2ob2ob2ob2o3bo3b2obobo2bob2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob
2obo$ob2ob2ob2ob2ob3ob2ob2ob2ob2ob2ob2ob2o$2ob2ob2ob2o3bo3b2obobo2bob
2ob2ob2ob2o$b2ob2ob2ob3o2b2o5b6ob2ob2ob2obo!
```

That might not quite fit with Achim Flammenkamp's definition of a Garden of Eden, but it fits pretty well with common usage. (Achim's term for a finite group of cells with no parent is "orphan", and you embed an orphan in an infinite field of OFF cells to get a GoE.)

It's certainly easy enough to describe the infinite background pattern that would constitute (most of) the parent for any finite pattern you might choose in an XOR rule. It just wouldn't be a finite parent.

-- Did I say all that right? It's so easy to get tripped up on definitions with these GoEs/orphans...

Posted: **January 4th, 2019, 4:41 pm**

Yeah, that what more or less what I said in my post (which I assume I published while you were already writing yours), plus an example of the sort I was looking for.

Posted: **January 4th, 2019, 5:05 pm**

When exactly was the glider discovered? The Glider article a quote from *Genius at Play* which says it was found in the fall of 1969 while investigating the R-pentomino, but just before the quote claims 1970. Other articles also support the 1970 date; the wiki pages for block and blinker both say they were found in 1970, and therefore the glider would have to have been discovered afterwards since they were already present in the R-pentomino's evolution.

Posted: **January 4th, 2019, 5:13 pm**

I think most sources (including the article infobox) say 1970 because that's when Life was first written about and made known to the general public (Gardner wrote his first article on Life in 1970). However, Conway and his gang were (of course) investigating Life beforehand. Given that Siobhan Roberts had access to Conway and interviewed him extensively for his biography, I would wager that 1969 is in fact correct. However, it's merely an "internal" figure; nothing was published (in any sense of the word), which is why 1970 is usually quoted.Ian07 wrote:When exactly was the glider discovered? The Glider article a quote fromGenius at Playwhich says it was found in the fall of 1969 while investigating the R-pentomino, but just before the quote claims 1970. Other articles also support the 1970 date; the wiki pages for block and blinker both say they were found in 1970, and therefore the glider would have to have been discovered afterwards since they were already present in the R-pentomino's evolution.

For much the same reason, other small patterns were discovered by Conway & Co. before 1970, and the 1970 figure again refers to Gardner's article.

As for the wiki, BTW, we never tried to find a consensus as to what dates should be given. I added the quote from Conway's biography since it was interesting, informative and apropos, but I did not want to go against long-established consensus that patterns only known to members of Conway's gang before 1970 would be considered to have been "discovered" with the publication of Gardner's first article.

Posted: **January 5th, 2019, 6:06 am**

It sounded right to me. But wouldn't it be much easier if the definitions were the other way around? To me "Garden" suggests a bounded region, whereas "orphan" just means "has no parents". Maybe we should just declare that they've swapped, especially since everyone talks about finding GoEs when they mean finding orphans.dvgrn wrote:-- Did I say all that right? It's so easy to get tripped up on definitions with these GoEs/orphans...

It would also be nice to have a term for a finite pattern with a parent but no finite parent, like the "GOE" pattern you posted. It's an open problem whether these exist in Life.

This makes sense. Since Gardner's article was published in October 1970, we have no way of knowing which patterns (except the glider) were found in 1969.Apple Bottom wrote:I did not want to go against long-established consensus that patterns only known to members of Conway's gang before 1970 would be considered to have been "discovered" with the publication of Gardner's first article.

Posted: **January 5th, 2019, 7:59 am**

It's also possibly worth saying that by using a 'proof by compactness' you can show (in any CA) that if every finite subset of an infinite universe has a parent (that is allowed to produce junk around the edges), then the entire universe has a (possibly infinite) parent. So the fact that every pattern has a (not necessarily finite) parent is not unique to XOR rules; it applies to every GoE-less rule.dvgrn wrote:It's certainly easy enough to describe the infinite background pattern that would constitute (most of) the parent for any finite pattern you might choose in an XOR rule. It just wouldn't be a finite parent.

-- Did I say all that right? It's so easy to get tripped up on definitions with these GoEs/orphans...

Posted: **January 5th, 2019, 9:35 am**

Yeah, I certainly worried a little bit that creating a "1969" category for pattern discoveries was opening an unnecessarily messy can of worms. If anyone wants to change the date back to 1970, it's fine by me.Macbi wrote:This makes sense. Since Gardner's article was published in October 1970, we have no way of knowing which patterns (except the glider) were found in 1969.Apple Bottom wrote:I did not want to go against long-established consensus that patterns only known to members of Conway's gang before 1970 would be considered to have been "discovered" with the publication of Gardner's first article.

This morning I added a "historical note" to the Glider article, which should explain the 1969/1970 ambiguity, whichever date is given in the infobox.

Posted: **January 5th, 2019, 10:47 am**

We know that the R-pentomino dates from before the glider (because the latter was discovered by simulating the R-pentomino). Also, I recall that the polyominoes were investigated in increasing order of size, and that the R was the last pentomino to be resolved. This implies that the block, blinker, beehive, traffic lights, and loaf were known by 1969:dvgrn wrote:Yeah, I certainly worried a little bit that creating a "1969" category for pattern discoveries was opening an unnecessarily messy can of worms. If anyone wants to change the date back to 1970, it's fine by me.Macbi wrote:This makes sense. Since Gardner's article was published in October 1970, we have no way of knowing which patterns (except the glider) were found in 1969.Apple Bottom wrote:I did not want to go against long-established consensus that patterns only known to members of Conway's gang before 1970 would be considered to have been "discovered" with the publication of Gardner's first article.

This morning I added a "historical note" to the Glider article, which should explain the 1969/1970 ambiguity, whichever date is given in the infobox.

Code: Select all

```
x = 108, y = 44, rule = B3/S23
81b3o$82bo2$41bobo$41b3o29b2o$73b2o$5o3$20b4o$21bo51b2o6b2o15bo$51b2o
20bo8b2o14bo7bo$52b2o19bo24bo7b2o$53bo5$77b4o2$29bo$29b2o72b2o$30bo$
30bo2$5b4o$5bo$104bo$44b2o$44bo$43b2o5$24bo$23b3o32b2o$24bo33b2o$58bo
3$11b3o26bo$12bo27bo$12bo27b3o!
```

Bill sent me an e-mail from Dick Esterle saying that Siobhan Roberts wrote:I'm afraid I tried to pin down Life dates and Conway was wary of doing so, and indeed could not himself, hence I told it like it was in the book. I'd say circa 1967/68. RKG might be able to provide something more specific.

Posted: **January 5th, 2019, 3:27 pm**

Uh-oh. Can of worms was right -- the R-pentomino has now migrated backwards to 1969 on the LifeWiki.calcyman wrote:We know that the R-pentomino dates from before the glider (because the latter was discovered by simulating the R-pentomino). Also, I recall that the polyominoes were investigated in increasing order of size, and that the R was the last pentomino to be resolved. This implies that the block, blinker, beehive, traffic lights, and loaf were known by 1969...

I'd

Also the B-heptomino, of course, but it was the not-really-named B-heptaplet form of the B, so maybe that doesn't count. Then again, quite possibly very few of these things besides the glider were named in 1969.

-- I guess that still seems like a reasonably small collection of objects that can be definitively moved to http://conwaylife.com/wiki/Category:Patterns_found_in_1969. Maybe we can just decide to do these, or some subset of them, and leave all the long canoes and snakes and barges and other probably-early-discovered things in 1970.

Posted: **January 6th, 2019, 7:01 am**

Another piece of evidence to support that is that R pentomino forms 3 blocks and a blinker before the first glider...

Posted: **January 12th, 2019, 4:46 am**

Has an SMS or SSS breeder been constructed?

What about it in other rules?

What about it in other rules?

Posted: **January 12th, 2019, 7:59 am**

I think I've seen one before. However I can't find the link.Saka wrote:Has an SMS or SSS breeder been constructed?

What about it in other rules?

Posted: **January 12th, 2019, 8:29 am**

Paul Tooke's pianola breeder experiments include something he labeled as an SSS breeder.Hunting wrote:I think I've seen one before. However I can't find the link.Saka wrote:Has an SMS or SSS breeder been constructed?

What about it in other rules?

The "super breeders?" topic talks about MSS, and here's an SMS breeder also by Paul Tooke.

Anyone want to decide where to put these links in the LifeWiki so that they don't get lost again?

Posted: **January 27th, 2019, 10:32 pm**

Here's something I've spent quite a bit of time trying to find now, so I'm hoping other people have better memories than I do:

1) Some time ago, maybe 2016-ish (?) someone posted a failed 2c/3 signal elbow, or maybe it was a failed 5c/9-to-2c/3 signal elbow. If I remember right, a couple of bits toward one edge didn't get restored. Seems like the kind of thing that might be worth trying a fresh LLS search on -- but now I can't find the relevant message!

Where was it? The only thing I can seem to find is Dean Hickerson's original failed 5c/9-to-2c/3 from 1997:

I'm kind of thinking that the new failed elbow was a hand-modified version of this one, but I could be wrong about that.

**Bonus question:**

2) Are there any relatively inexpensive ways to restore the damage to this glider-to-2c/3 converter?

1) Some time ago, maybe 2016-ish (?) someone posted a failed 2c/3 signal elbow, or maybe it was a failed 5c/9-to-2c/3 signal elbow. If I remember right, a couple of bits toward one edge didn't get restored. Seems like the kind of thing that might be worth trying a fresh LLS search on -- but now I can't find the relevant message!

Where was it? The only thing I can seem to find is Dean Hickerson's original failed 5c/9-to-2c/3 from 1997:

Code: Select all

```
x = 63, y = 40, rule = LifeHistory
4.A.2A$4.2A2.A$7.A2.A$2.5A.2A.A2.A$.A4.A.A2.4A$.A2.A3.3A6.A$2A.A.3A2.
A2.5A38.2A$3.A6.A.A5.2A35.A2.A2.2A$2A.A.3A2.A2.A.2A.A.A34.A.A.A2.A$A
2.A.A2.A.2A.A.A.A2.A31.2A.A.A2.2A$2.2A2.A2.A3.A.A4.A.2A29.A.A.A$4.2A
4.4A.2A2.2A2.A29.A.A2.4A$4.A3.A.A6.A3.A28.2A.A.A5.A$5.4A.A.5A.3A3.A
24.A.A.A2.3A$9.A.A4.A.A2.4A24.A.A.A4.A$7.A3.A2.A3.A.A6.A18.2A.A.A2.4A
$7.2A2.A.A.4A.A2.5A19.A.A.A$10.2A.A6.A.A5.2A3.2A12.A.A2.4A$13.A.4A.A
2.A.2A.A.A.A.A9.2A.A.A5.A$13.A.A2.A.2A.A.A.A2.A.A10.A.A.A2.3A$12.2A2.
A2.A3.A.A4.A.2A9.A.A.A4.A$14.2A4.4A.2A2.2A4.A4.2A.A.A2.4A$14.A3.A.A6.
A3.4A.A4.A.A.A$15.4A.A.5A.3A5.A4.A.A2.4A$19.A.A4.A.A2.6A.2A.A.A5.A$
17.A3.A2.A3.A.A6.A.A.A2.3A$17.2A2.A.A.4A.A2.2A2.A.A.A4.A$20.2A.A6.A.A
2.A.A.A2.4A$23.A.4A.A2.A.A.A.A$23.A.A2.A.2A.A.A.A2.4A$20.A.2A2.A2.A3.
A.A.A5.A$20.2A2.2A4.AC2A.A2.3A$23.A2.4A3.D.A4.A$24.A5.2ACA.C4A$25.3A
2.A2.A$27.A3.A2.6A$28.3A5.A2.A2$28.2A.A$28.A.2A!
```

2) Are there any relatively inexpensive ways to restore the damage to this glider-to-2c/3 converter?

Code: Select all

```
#C A glider hits a still-life, sending a 2c/3 signal along a diagonal.
#C Another glider on the same path causes the pattern to decay.
#C Dean Hickerson, 4/11/97
x = 132, y = 133, rule = LifeHistory
.A$2.A6.A$3A5.A.A$8.A.A$7.2A2.2A$4.2A3.A.A$4.A.4A2.A$10.A.A$4.A.5A.2A
3.A$2.3A.A8.3A$.A4.A.7A$2.3A.A.A6.A$4.A.A.A2.5A$7.2A.A7.A$10.A2.6A$
10.A.A$9.2A.A2.6A$12.A.A6.A$12.A.A2.5A$13.2A.A7.A$16.A2.6A$16.A.A$15.
2A.A2.6A$18.A.A6.A$18.A.A2.5A$19.2A.A7.A$22.A2.6A$22.A.A$21.2A.A2.6A$
24.A.A6.A$24.A.A2.5A$25.2A.A7.A$28.A2.6A$28.A.A$27.2A.A2.6A$30.A.A6.A
$30.A.A2.5A$31.2A.A7.A$34.A2.6A$34.A.A$33.2A.A2.6A$36.A.A6.A$36.A.A2.
5A$37.2A.A7.A$40.A2.6A$40.A.A$39.2A.A2.6A$42.A.A6.A$42.A.A2.5A$43.2A.
A7.A$46.A2.6A$46.A.A$45.2A.A2.6A$48.A.A6.A$48.A.A2.5A$49.2A.A7.A$52.A
2.6A$52.A.A$51.2A.A2.6A$54.A.A6.A$54.A.A2.5A$55.2A.A7.A$58.A2.6A$58.A
.A$57.2A.A2.6A$60.A.A6.A$60.A.A2.5A$61.2A.A7.A$64.A2.6A$64.A.A$63.2A.
A2.6A$66.A.A6.A$66.A.A2.5A$67.2A.A7.A$70.A2.6A$70.A.A$69.2A.A2.6A$72.
A.A6.A$72.A.A2.5A$73.2A.A7.A$76.A2.6A$76.A.A$75.2A.A2.6A$78.A.A6.A$
78.A.A2.5A$79.2A.A7.A$82.A2.6A$82.A.A$81.2A.A2.6A$84.A.A6.A$84.A.A2.
5A$85.2A.A7.A$88.A2.6A$88.A.A$87.2A.A2.6A$90.A.A6.A$90.A.A2.5A$91.2A.
A7.A$94.A2.6A$94.A.A$93.2A.A2.6A$96.A.A6.A$96.A.A2.5A$97.2A.A7.A$100.
A2.6A$100.A.A$99.2A.A2.6A$102.A.A6.A$102.A.A2.5A$103.2A.A7.A$106.A2.
6A$106.A.A$105.2A.A2.6A$108.A.A6.A$108.A.A2.5A$109.2A.A7.A$112.A2.6A$
112.A.A$111.2A.A2.6A$114.A.A6.A$114.A.A2.5A$115.2A.A7.A$118.A2.6A$
118.A.A8.A$117.2A.A2.7A$120.A.A$120.A.A2.5A$121.2A.A4.A$124.A2.A$124.
A.A.4A$123.2A.A4.A$127.3A$129.2A!
```

Posted: **January 28th, 2019, 3:05 pm**

How do you search for catalysts on a Mac?

Posted: **January 29th, 2019, 12:36 pm**

Are there any glider eaters consisting entirely of blocks that meet these conditions?

1) The eater, when fully recovered, consists entirely of blocks and pseudo/quasi still lives that consist of blocks. Biblock is valid, while beehive is not, for example.

2) The eater recovers quickly, relative to its bounding box. In other words, approximately <= the time it takes for a c/4o spaceship to reach the far end, but a little longer is acceptable.

3) All blocks must play some role in the eating reaction. You cannot put a block a billion cells away from the others and claim that the eater is fast considering its bounding box.

4) Having a universal constructor rebuilding the blocks is not a valid solution, for the sake of the argument, unless they are unusually fast (no more than 1000 gens recovery time)

5)The solution should preferably me small enough to run in lifeviewer. In other words, it should be smaller than the minstrel detector and remover. This condition does not need to be satisfied.

1) The eater, when fully recovered, consists entirely of blocks and pseudo/quasi still lives that consist of blocks. Biblock is valid, while beehive is not, for example.

2) The eater recovers quickly, relative to its bounding box. In other words, approximately <= the time it takes for a c/4o spaceship to reach the far end, but a little longer is acceptable.

3) All blocks must play some role in the eating reaction. You cannot put a block a billion cells away from the others and claim that the eater is fast considering its bounding box.

4) Having a universal constructor rebuilding the blocks is not a valid solution, for the sake of the argument, unless they are unusually fast (no more than 1000 gens recovery time)

5)The solution should preferably me small enough to run in lifeviewer. In other words, it should be smaller than the minstrel detector and remover. This condition does not need to be satisfied.

Posted: **January 29th, 2019, 2:45 pm**

No Blockic glider eaters or glider-to-anything converters are known, given your restriction #4.Moosey wrote:Are there any glider eaters consisting entirely of blocks that meet these conditions?...

4) Having a universal constructor rebuilding the blocks is not a valid solution, for the sake of the argument, unless they are unusually fast (no more than 1000 gens recovery time)

Even without restriction #4, nobody has actually built anything that fills the bill, and I don't think anybody knows a way to do it that wouldn't also violate your restriction #2.

The problem is that we don't have enough blocks-only Herschel conduits to make a universal set. We have R64 and B60 and whatever-that-other-elementary-conduit-is, but they don't connect to each other at all, let alone in a way that would allow for splitting off signals and repairing circuits that only work once.

Works-once Blockic conduits are actually pretty common. For example, here's a glider-to-Herschel converter based on Paul Callahan's receiver:

Code: Select all

```
x = 55, y = 66, rule = LifeHistory
21.2A$21.2A3$19.2A$19.2A3$8.2C$8.2C$4.2A$4.2A7$17.2A$17.2A29.2A$48.2A
3$35.2C$35.2C2$45.2C$45.2C3$.A8.2A$.2A7.2A$A.A28$48.2C$48.2C3$53.2C$
53.2C!
```

Really we wouldn't do it that way, though. It would probably be better to hunt for a Blockic constellation that produced a clean output Herschel or two, and then use hypothetical Blockic splitters to rebuild that constellation directly. But none of this can work until someone finds more Blockic conduits.

Posted: **January 29th, 2019, 3:43 pm**

What about one where restriction 2 is removed?dvgrn wrote:No Blockic glider eaters or glider-to-anything converters are known, given your restriction #4.Moosey wrote:Are there any glider eaters consisting entirely of blocks that meet these conditions?...

4) Having a universal constructor rebuilding the blocks is not a valid solution, for the sake of the argument, unless they are unusually fast (no more than 1000 gens recovery time)

Even without restriction #4, nobody has actually built anything that fills the bill, and I don't think anybody knows a way to do it that wouldn't also violate your restriction #2.

The problem is that we don't have enough blocks-only Herschel conduits to make a universal set. We have R64 and B60 and whatever-that-other-elementary-conduit-is, but they don't connect to each other at all, let alone in a way that would allow for splitting off signals and repairing circuits that only work once.

Works-once Blockic conduits are actually pretty common. For example, here's a glider-to-Herschel converter based on Paul Callahan's receiver:

If we had enough Blockic conduits to make a universal set (including a signal splitter) then we could use this to get a Herschel, and then produce gliders to rebuild the five blocks that go missing when the glider comes in -- and then we'd have a Blockic glider eater.Code: Select all

`x = 55, y = 66, rule = LifeHistory partial blockic conduit`

Really we wouldn't do it that way, though. It would probably be better to hunt for a Blockic constellation that produced a clean output Herschel or two, and then use hypothetical Blockic splitters to rebuild that constellation directly. But none of this can work until someone finds more Blockic conduits.

Or what if, instead, beehives, and loaves were also allowed?

Posted: **January 29th, 2019, 4:35 pm**

Makes no difference. You have to either lift restriction #4 or also allow boats (don't need beehives so much). The first option lets you theoretically design some UC-based system with a stage where all the required data is encoded into blocks. Don't go there; it wouldn't be much fun at all to try to build such a ridiculous thing in practice, without some new technological breakthroughs anyway.Moosey wrote:What about one where restriction 2 is removed?

Or what if, instead, beehives, and loaves were also allowed?

The second option gives you access to the sidesnagger, so you'd have a glider eater that answers your original question.

To get a universal set of Herschel conduits, even blocks, boats, beehives and loaves aren't enough -- you really need fishhook eaters to get anywhere. But of course that would trivially answer your eater question with no need for any crazy complicated bait or staged recovery mechanisms.

Posted: **January 29th, 2019, 4:37 pm**

Okay, so if restrictions 2 and 4 are removed... What's the smallest universal constructor that would do that?dvgrn wrote:Makes no difference. You have to either lift restriction #4 or also allow boats (instead of beehives) -- that gives you access to the sidesnagger, so you have a glider eater that answers your original question.Moosey wrote:What about one where restriction 2 is removed?

Or what if, instead, beehives, and loaves were also allowed?

To get a universal set of Herschel conduits, even blocks, boats, beehives and loaves aren't enough -- you really need fishhook eaters to get anywhere. But of course that would trivially answer your eater question with no need for any crazy complicated bait or staged recovery mechanisms.

Posted: **January 29th, 2019, 5:08 pm**

It's fairly painful. There's a description by calcyman "(where x = (0,0))" that probably comes pretty close. You can get a vague sense of how big something like this would be, by looking at half of the Caterloopillar.Moosey wrote:Okay, so if restrictions 2 and 4 are removed... What's the smallest universal constructor that would do that?

The Caterloopillar uses loaves to store data instead of blocks, but maybe we could figure something out with blocks instead, where the blocks get regenerated on the same spot and a glider is produced. The Caterloopillar stores enough data to build seeds for one small salvo of *WSSes, whereas we might need multiple parallel lines of blocks and "reader" *WSS salvos to make the slow glider salvos that we'll need to bend a construction arm around a couple of elbows and rebuild the *WSS salvo seeds that started the whole thing off.

-- I'm not really too happy with my mental picture of how this will work, when I get down into the details. There's probably something slightly simpler available these days. But it's definitely Not Small, and not trivial to actually build it... and when you're all done, the darn thing is just an eater. (!)