Talk:Full-employment theorem

Google reports 45 hits for this phrase, including at Wikipedia. The article seems to bear scant relation to the information given in the articles Google lists. Lots of typos etc.--duncan 21:08, 12 July 2006 (UTC)


 * 62 now .--MarSch 13:30, 24 December 2006 (UTC)

It's used in Wikipedia in other places with this meaning. Perhaps the page needs to go onto Wikitionary?

Watson Ladd 02:00, 30 December 2006 (UTC)


 * 8550 results as of today. linas (talk) 02:47, 6 June 2012 (UTC)

nonsense
This stuff is bogus. Just because the set of size-optimizing compilers does not have a most-efficient member, doesn't mean that _humans_ must always write better size-optimizing compilers. Computers can do this much better, once we get some decent AI. Thus I have to call "bullshit!". --MarSch 17:29, 18 March 2007 (UTC)


 * Actually reading less non-carefully I see that the article states this already. --MarSch 17:35, 18 March 2007 (UTC)


 * Anyway there can never be such a thing as a full employment theorem as defined by this article. I'll try prodding. --MarSch 17:37, 18 March 2007 (UTC)

I've removed this article from proposed deletion as, regardless of the question of what a full employment theorem would guarantee in the presence of AI, the term full employment theorem is in fact used in the sense given in this article. (Also, as the proposer acknowledges, the article already treats the question of whether a full employment theorem would guarantee full employment in the presence of AI.) The proposer's claim that there are no full employment theorems which satisfy the definition of the article is not correct. Spacepotato 22:24, 21 March 2007 (UTC)


 * Did you know that however decent some AI is, it will not solve the halting problem? Neither do humans. —Preceding unsigned comment added by 90.191.147.186 (talk) 21:25, 5 July 2010 (UTC)


 * Actually it is not nonsense. You haven't studied computability, and saying "once we get some decent AI" assumes strong AI, which is itself a subject of debate with little evidence on either side. If one is restricting "AI" to be run on a conventional (turing equivalent) computer, then the AI is a formal system is subject to incompleteness as the article claims.  Whether some non-computer form of AI can be developed is unknown.  Perhaps a better way of attacking the full employment theorem would be to point out that it is not known whether humans are equivalent to formal systems (which would require that we are deterministic, for example).  Penrose famously argues no, but his argument is pretty speculative.  —Preceding unsigned comment added by 122.59.233.21 (talk) 16:26, 15 July 2010 (UTC)


 * This "theorem" is flippant but not bullshit or nonsense. The readers who say so would do well to study incompleteness, church-turing, and the debate about whether these apply to humans.  In any case, it is exactly the point that they apply to computers as they are currently (and have generally been) understood, so AI (as implemented in a computer) will not help.  In fact a variation of the full employment theorem would say that human brains will be employable despite AI, precisely because of incompleteness.  —Preceding unsigned comment added by 89.187.142.72 (talk) 05:55, 25 September 2010 (UTC)

out of context
the problem of this article is that it is out of context. the term is somewhat humorous, and the article would better refer to that, instead of speculate about AI (??), etc. --187.40.167.205 (talk) 08:52, 29 August 2010 (UTC)


 * yes, I think I fixed it. Bhny (talk) 19:47, 30 April 2014 (UTC)


 * Bhny, while I think this article has a few subtle problems, I don't think the edits you made were in the completely right direction — Preceding unsigned comment added by 104.7.96.38 (talk) 18:15, 4 May 2014 (UTC)

wrong statement
"Thus, the existence of a provably perfect size-optimizing compiler would imply a solution to the halting problem, which cannot exist, making the proof itself an undecidable problem." The highlighted conclusion seems wrong. The problem (like the halting problem) is undecidable, but proving the undecidability is not. In fact the article outlines a proof via the halting problem a few sentences earlier. — Preceding unsigned comment added by 46.223.103.62 (talk) 10:49, 27 February 2018 (UTC)
 * I've fixed this. TripleShortOfACycle (talk) 03:01, 14 February 2020 (UTC)