Talk:algorithm

I deliberately left out "finite". As used here it's ambiguous. It could mean
 * 1) An algorithm has a finite description, which is covered by "well-defined"
 * 2) An algorithm must terminate after a finite number of steps.

While finite termination is sometimes given as a criterion, it turns out to be more trouble than it's worth. For example, the standard algorithm for implementing a server is essentially an infinite loop:

forever get request process request

or more completely

while not terminated get request process request (may set terminated)

In the abstract, neither of these is guaranteed to terminate in a finite number of steps, though in practice entropy will win in the end. For further examples, google "algorithm does not terminate". -dmh

(P.S.: What language was the name of the al-Khwarizmi in, if not Arabic :-)


 * OK, fair points, but the mathematical definition of an algorithm is one that has a finite number of steps (Wikipedia and dictionary.com agree with me, although Mathworld does not - I will check my mathematical dictionaries at home when I can). An procedure that does not terminate cannot have completed its task, and so cannot be an algorithm.  So I would say that "finite" is a crucial part of the definition and should stay in.


 * Of course, the definition of the halting problem implicitly says that an algorithm need not terminate (otherwise it would not use the word "algorithm". Maybe we should add "strictly" or "usually" to qualify this.  It's clearly a moot point.


 * Al-Khwarizmi's name was Arabic, of course, but the etymology is from the English transliteration rather than from the Arabic form. I'll check the OED when I remember to.


 * &mdash; Paul G 12:25, 9 Nov 2004 (UTC)


 * It's too late for me to check right now but I was pretty sure Al-Khwarizmi was Persian. This may well be an Arabicized version of his name though, which was a pretty common thing. &mdash; Hippietrail 14:13, 9 Nov 2004 (UTC)


 * Persian sounds right. Paul, is there a particular reason you want to classify this as a "corruption"?  Unless there's some technical linquistic sense of the word I'm not aware of, this just seems to place a value judgement on the normal process of words chagning in the process of borrowing.  I'm not going to change it back, but I don't see a good reason for it.


 * As to finite, I'll add a usage note. One of the google hits I found is a good case in point.  It talks about a network algorithm going into an infinite loop under some presumably pathological condition.  I got the "finite" part in high school, but as a software engineer I can say for sure that no one really worries about it.  Some algorithms do great in their comfort zone but misbehave on bad input.  As a practical matter, you either guard the input or choose a more robust algorithm.  For that matter, algorithms that terminate in the technical sense can take practically infinite time to execute. -dmh 15:02, 9 Nov 2004 (UTC)


 * "Corruption" is a linguistic term that I have seen used in other dictionaries, but I admit it sounds like a value-judgement. dictionary.com defines this meaning as "The act of changing, or of being changed, for the worse; departure from what is pure, simple, or correct; as, a corruption of style; corruption in language." so there is value-judgement in there.  I'll change it to something better (see further comments about the etymology below).


 * The sources I checked earlier today vary on whether they include "finite" or not. Collins Reference Dictionary of Mathematics (CRDM), which is quite thorough and precise in its definitions, excludes it; the Encyclopedic Dictionary of Mathematics (2nd ed.), which is very comprehensive, describes an algorithm as a finitary procedure for carrying out a task, where "finitary" means (according to CRDM) "not involving infinite sets, either explicitly or implicitly".  This would cover both possible interpretations of "finite" here - a procedure with a finite number of steps, and a procedure that takes a finite length of time (ie, that terminates).  The OED does not mention the word "finite" in its definition.


 * Given the variation, I think a usage note is appropriate in this case.


 * As for the etymology, the OED's is more extensive, saying that it the word comes from (I summarise there - the OED gives much more) the Old French "algorisme", which comes from the med(ieval?) Latin "algorismus", which comes from the Arabic surname of the mathematician, which it spells as "al-Khowarazmi"; the word has been influenced by the (ancient) Greek "άριθμός" (the accent on the alpha should be grave, but I don't have that to hand), meaning "number". So  a suitable summarised etymology might be something like "ultimately from al-Khowarazmi" (or however we want to spell it).


 * &mdash; Paul G 19:14, 10 Nov 2004 (UTC)


 * I've changed "corruption" to "modification". &mdash; Paul G 19:16, 10 Nov 2004 (UTC)

RFD discussion: August 2018
(Scottish) I don't see the sense in having a Scottish entry. Even the brief Wikipedia article is mostly in standard English. DonnanZ (talk) 09:53, 18 August 2018 (UTC)
 * Keep, because we don't delete entries just because you don't like them. You can send it to RFV if you don't think it's attested. —Μετάknowledge discuss/deeds 21:00, 19 August 2018 (UTC)
 * Keep. We treat Scots as a language, not as a subset of English. — Ungoliant (falai) 21:41, 19 August 2018 (UTC)


 * Transferred to RFV Non-English. DonnanZ (talk) 08:38, 21 August 2018 (UTC)

RFV discussion: August 2018–April 2019
(Scottish) I don't see the sense in having a Scottish entry. Even the brief Wikipedia article is mostly in standard English. DonnanZ (talk) 09:53, 18 August 2018 (UTC)
 * Keep, because we don't delete entries just because you don't like them. You can send it to RFV if you don't think it's attested. —Μετάknowledge discuss/deeds 21:00, 19 August 2018 (UTC)
 * Keep. We treat Scots as a language, not as a subset of English. — Ungoliant (falai) 21:41, 19 August 2018 (UTC)


 * Transferred from RFD Non-English. DonnanZ (talk) 08:33, 21 August 2018 (UTC)
 * Best I could find so far: "Ye micht write it as a (kynd o) algorithm:" (clearly Scots, but from a listserv group). I don't think this entry is going to make it. ←₰-→ Lingo Bingo Dingo (talk)  07:51, 22 August 2018 (UTC)
 * RFV failed. — Mnemosientje (t · c) 14:49, 11 April 2019 (UTC)

"The Algorithm"
I think it may be worth noting that from the 2010s on, the word, especially when used with a definite article, has frequently been used in non-technical contexts to refer to content ranking / recommendation systems used by social media websites (e.g., "the Twitter algorithm"). This is technically just a specific case of the more general definition of algorithm, of course. Any thoughts? 70.172.194.25 03:37, 29 May 2022 (UTC)