Talk:Super-recursive algorithm

This article was nominated for deletion on 25 May 2008. The result of the discussion was no consensus.

Archives

Copyediting 2009-03-14

I did some copyediting today, after someone wisely pointed out that the article was very difficult to read.

At present, I am pretty happy with the neutrality of the lede section. The next few sections are just OK, and the section "Schmidhuber's generalized Turing machines" is pretty rough. — Carl (CBM · talk) 00:13, 15 March 2009 (UTC)[reply]

This page is still here?

Amazing.

The page is misleading on (at least) two points. The first is the content, but it is not sufficient to be "neutral" with respect to that. Because the second is the term itself. There's room on Wikipedia for notable dissent with the Church-Turing thesis, but "Super-recursive algorithm" is still an idiosyncratic term, and most of the dissenters cited by Burgin do not use it.

I still advocate very heavily trimming this page and stuffing whatever's left into hypercomputation or other target. --Unzerlegbarkeit (talk) 05:33, 28 December 2009 (UTC)[reply]

I have started. Is there someone who can explain the difference between "inductive turing machine" and Schmidhuber's scheme? --Unzerlegbarkeit (talk) 00:43, 1 August 2010 (UTC)[reply]

I can sympathize. I tried to get this article thrown out on the grounds that there is no single, independent, peer-reviewed article specifically about "super-recursive algorithms". I doubt that Multipundit could produce one. Yakushima (talk) 01:51, 23 December 2010 (UTC)[reply]

Since there is an article hypercomputation, which discusses the topic and mentions its different incarnations, I agree that there is no room for a separate article. It should be merged. Is it possible to restart a merge discussion (instead of deletion)? I do not know Wikipedia procedures well enough. AmirOnWiki (talk) 14:02, 14 October 2011 (UTC)[reply]

Correct me if I'm wrong, but by the definition listed here (the result simply has to be reached "at some point") couldn't you just enumerate all numbers and claim that you had a computer that eventually solved every problem with a numeric solution? I assume there must be something I'm missing because a claim that this represents any sort of useful definition of computability is baffling to me. 173.79.253.74 (talk) 18:36, 11 July 2013 (UTC)[reply]

I do not think you are missing anything. This Burgin person seems to be fundamentally confused - since these "inductive" Turing machines either do not halt or else could be trivially simulated by ordinary Turing machines, they do not allow you to calculate anything a Turing machine can't, and therefore don't contribute anything new to the field. Similar for genetic computing and fuzzy logic computing - plenty of those have been written in normal computing languages and run on normal computers, and are therefore simulatable by Turing machines and not anything new. The article therefore does not contribute anything useful to an encyclopaedic description of the field. If we have to have an article about his book, make it about his book, not purporting to be an article about the study of algorithms. 116.199.214.34 (talk) 03:45, 27 November 2013 (UTC)[reply]

inadequate explanation

Does the article explain the concept to viewers not familiar with it already?

Is the information providing enough context to mean anything without referring to the book mentioned?

2001:470:600D:DEAD:E94B:92C8:B4E1:F8C5 (talk) 11:06, 6 August 2014 (UTC)[reply]

Merge with hypercomputation?

This is what was proposed in the AfD by Carl: Wikipedia:Articles for deletion/Super-recursive algorithm. Should we do this? One issue is that there are an awful lot of citations in this article that may be difficult to house in the hypercomputation article while keeping balance. — Charles Stewart (talk) 15:21, 2 May 2016 (UTC)[reply]

A merge sounds good to me. Most of the references are not explicitly cited and would not be strictly necessary to maintain during the merge; the number of inline citations is smaller and I think would be manageable. — Carl (CBM · talk) 17:18, 2 May 2016 (UTC)[reply]

A preprint by Burgin that might clear up some of the confusion about Burgin's writing

I've read most of the older Deletion request, and the talk page here. I found this Wikipedia page due to having recently found a preprint by Burgin while investigating something. I suspect that the majority of complaints about the unaccessability of Burgin's work could get become more clear to editors well versed in Computer Science after reading it, as it links Burgin's conception of 'superrecursiveness' with Interactive Computation. Preprint here:

https://arxiv.org/abs/0710.1455 - Superrecursive Features of Interactive Computation

But, unfortunately, it has the same peer review & Runglish issues others have remarked upon.

--No identd (talk) 18:45, 17 June 2018 (UTC)[reply]

I don't think I have seen many complaints that Burgin's work was unaccessible - the main criticisms in print have been that his concept of "algorithm" is both well known already as limit computability, and also not an actual "computability" because there is no way to tell what answer is being presented. The same thing happens for interactive computation, which is simply a kind of oracle computation, which was studied by Kleene. — Carl (CBM · talk) 00:16, 18 June 2018 (UTC)[reply]

Turing machines cannot edit their previous outputs?

"Turing machines cannot edit their previous outputs" - What is that supposed to mean? The interpretation that is most obvious to me ("once a symbol is written on the tape, it can never be changed") is clearly wrong. Chrisahn (talk) 03:57, 23 June 2018 (UTC)[reply]

There are two ways to interpret this. One traditional model of Turing machine has a write-only output tape, which cannot be edited once it is written. They also have a work tape, of course, this is just an output convention. The second way to interpret it is that, once a Turing machine computes e.g. f(1), it cannot go back when computing f(2) to change its mind about the value that f(1) should have had. In the models being discussed in that section, the machine can give a definite answer to a question, but later revise that answer to something else.

Either way, what the section is describing is a machine for performing a limit computation: the machine occasionally writes down values for each entry of an infinite sequence, but it can go back and change those values at any time. The machine has to eventually stop changing each value, so that in the limit there is a well defined output from the process, although the result after any finite number of steps may not reveal anything about that limit. — Carl (CBM · talk) 15:50, 23 June 2018 (UTC)[reply]

Thanks. Your explanation here is quite clear. I think the explanation in the section Schmidhuber's generalized Turing machines is still unclear. How about copying your explanation from here to the article? Maybe it's a bit long, but I think it would improve the text. Anyway, I don't care much either way. Chrisahn (talk) 22:24, 25 June 2018 (UTC)[reply]

Weird slab of OR - doing a reference cleanup

Somehow this pile of OR survived AFD many years ago. Anything that isn't nonsense should be merged into Hypercomputation.

The first thing to do is remove everything that's cited only to a preprint or isn't cited at all. Then we can see what's left - David Gerard (talk) 22:09, 6 February 2024 (UTC)[reply]

I cut a large amount of guff - especially the many attempts to claim previous works about hypercomputation, from before the Burgin book, for the term retrospectively. This article is specifically about this new Burgin thing.

I've also stressed that all of this is hypothetical, and if Burgin wants to disprove Church-Turing then this would turn mathematics on its head to the point that he first has to convince other computer scientists - David Gerard (talk) 22:18, 6 February 2024 (UTC)[reply]

I've dredged through Google Scholar. The term "super-recursive algorithm" sees very little usage, almost all of it from Burgin. I can see a few cites to it. This doesn't appear to be a notable book or academic term, or a concept in non-negligible usage. Even the Davis objection is a single paragraph saying that this is nonsense.

This should really just be redirected to Hypercomputation. Any objections? - David Gerard (talk) 15:15, 11 February 2024 (UTC)[reply]

Yes, please redirect. This article, as currently written, never says what a "super-recursive algorithm" is. I also examined older versions of this article, and none of them actually stated what this is, either. It seems to be something left to the imagination of the reader. 84.15.177.191 (talk) 15:09, 12 June 2024 (UTC)[reply]

Also, it appears that Named set theory is some OR ascribed to Burgin, and created by P. Zellweger aka User:Hpzell -- the only reference to this idea is a LinkedIn post by Zellweger promoting a database product. 84.15.177.191 (talk) 17:26, 12 June 2024 (UTC)[reply]