Talk:Algorithm

Where to cover "algorithms" as used in discussions about social media, big tech, etc?

I think this article should have some reference to the fact that "algorithms" are now being discussed in the context of social media platforms, "Big Tech", and related Internet technologies. As governments in Europe, the US, and other regions are discussing whether to regulate social platforms and companies such as Twitter, Facebook, and Google, the discussion often comes back to "algorithms". For instance, this is coming up quite often in discussions around Section 230 in the US. In Europe there is an EU activity of "Algorithm Awareness-Building." Just recently, two representatives introduced the "Protecting Americans from Dangerous Algorithms Act".

Given that many people may turn to Wikipedia to help understand what an "algorithm" is, I feel like there should be some mention of this usage on the page. But given the amount of detailed and academic info on the current page, I'm not sure how to best integrate this other content. Perhaps something brief in the lead paragraph and then a mention under "Informal definition"? Or a new section about "Algorithms in current politics"? Or a new section under "History" (although that seems mostly about the refinement of algorithms)? Any thoughts? - Dyork (talk) 17:25, 20 October 2020 (UTC)

I agree that there should be some mention of the social media usage on this page. However I think it's wrong to conflate that usage with the real meaning of the word "algorithm". In the social-media context, "algorithm" usually just means that some kind of processing has been done; it doesn't mean that the processing was in accordance with any kind of clear procedure, which is what an algorithm is supposed to be. I don't want this page to be about any old data-processing. That's not an algorithm. That's just what journos refer to as "algorithm". If we let that go, then we'll need a new term to refer to real algorithms: e.g. "Explicit procedure for performing transformations on data", which is a bit unwieldy.

Can we have our word back please, mister?

12:30, 27 October 2021 (UTC)

Absolutely. I'm surprised there isn't a separate article Algorithm (social media) that's written for a much more lay audience. Social media algorithms are usually more complex and aren't talked about in terms of big-O or anything like that. Instead, it's on the social impact of various choices of what to hilight -- things like the filter bubble, algorithmic radicalization, and even allegations that some social media company's decisions can sway political elections. And there are many news articles about this topic. --Hirsutism (talk) 18:45, 24 August 2021 (UTC)

Or perhaps an article about the Social impact of algorithms, with a heavy emphasis on social media algorithms? Because there are articles about algorithms being used to hire and fire people, for insurance and policing, and several other places they can introduce bias, and those should probably also be addressed in the for-a-lay-audience article. --Hirsutism (talk) 20:18, 24 August 2021 (UTC)

"In the social-media context, "algorithm" usually just means that some kind of processing has been done; it doesn't mean that the processing was in accordance with any kind of clear procedure, which is what an algorithm is supposed to be. I don't want this page to be about any old data-processing. That's not an algorithm. That's just what journos refer to as "algorithm". If we let that go, then we'll need a new term to refer to real algorithms: e.g. "Explicit procedure for performing transformations on data", which is a bit unwieldy."

Thank you! I've been trying to understand what news/social media means by algorithms. But none of the actual definitions help. Now I see that that is because the popular use of algorithm is similar to the popular use of 'metaphysics.' It means something different and sillier.

We Social impact of algorithms: I see that there is already a wiki dealing with Algorithmic Bias that probably accounts for this. Jesseraekern (talk) 18:14, 2 December 2022 (UTC)

I've written a bit of an article at User:Hirsutism/Algorithm (social media). Could someone take a look at it, and tell me if they think the article should be expanded to cover all AI-with-a-social-impact, or should it focus only on social media algorithms? On the other side, there's quite a few articles about this already, I don't know whether we even need a new separate article, other than a short summary in this article? --Hirsutism (talk) 20:41, 29 August 2021 (UTC)

References

The "algorithms" that are discussed in the context of social media are usually not algorithms as discussed in this article; the social media "algorithms" that are derived from big datasets by training ML systems are not transparent, and can't even really be expressed. I think it's incorrect to call these systems "algorithms".

However I do appreciate that journalists and ML promoters do refer to these systems as "algorithms". Therefore I support the creation of a page for "algorithms" in a social media context. That might make it easier to keep this page focused on the subject of real algorithms.

MrDemeanour (talk) 08:56, 15 September 2021 (UTC)

Agreed: "algorithm" as used in a social-media context is an important concept, quite distinct from this article's focus. The content of Hirsutism's draft at User:Hirsutism/Algorithm (social media) looks like a good start, but I'd argue that social media algorithms are simply a special case of a recommender systems, both on the technical and on the social/behavioral side. The social and behavioral aspects of recommender systems are mostly covered in the filter bubble article, but that's only one aspect. I'm not sure how to organize this better.... --Macrakis (talk) 14:14, 27 October 2021 (UTC)

'Algorithm' as used by non-computer-scientists / non-experts to discuss recommender systems in social media (which, surely, are implemented by large numbers of algorithms) is a technically-incorrect use of the word almost entirely unrelated to this article. — Preceding unsigned comment added by 72.200.126.73 (talk) 12:52, 23 August 2022 (UTC)

Page tidy up needed.

This page is quite long and rambling and seems to have some content that could go to other pages such as the section on Euclid's algorithm and the conversation between Turing etc. This is a gateway concept to many related issues and it would be good if it was easier for the general reader to navigate. I can get started on some of this. Amanda Lawrence 01:16, 27 October 2021 (UTC)

I whole-heartedly agree that this article strays off topic in many places. In particular, there is a paragraph in the "Expressing algorithms" section that talks about Turing machines. I have no problem with that idea, but it seems to come out of nowhere to me as currently written. I would either delete it or add a transitional phrase to the beginning of the first sentence. But I'm not going to do that right now, since there is no need to fine-tune this section if it is likely to be significantly modified in line with some of the other discussion on this Talk page.

Mike-c-in-mv (talk) 23:44, 18 June 2023 (UTC)

@AmandaSLawrence 117.20.112.19 (talk) 00:25, 6 January 2024 (UTC)

The whole section on Euclid's algorithm apart from the first para could be cut which would really help the flow - the examples could be quite short if they have their own page. But I'm not sure if all that long text should be added to the Euclid page or just cut. Amanda Lawrence 11:12, 25 January 2024 (UTC)

March 24, 2023

As added by User:2601:196:180:dc0:7c5a:37cb:a8c6:32d7:

:"One box is to be singled out and called the starting point. ...<!-- Is this punctuation (a period, followed by a space, then an elipse) correct? A recent editor assumed it was supposed to be this .... and changed it thusly. Can someone with access to the original source please verify it, and amend this inline note "stet" to indicate it should remain as verified.--> a specific problem is to be given in symbolic form by a finite number of boxes [i.e., INPUT] being marked with a stroke. Likewise, the answer [i.e., OUTPUT] is to be given in symbolic form by such a configuration of marked boxes...

Someone, i guess^{(talk i guess|le edit list)}

02:21, 25 March 2023 (UTC)

I think it's the end of the sentence, with text from the quote cut out. As in "Endofquotesentence. ...because reasons." with part of the quote removed. --Roundish ⋆tc) 02:35, 25 March 2023 (UTC)

Exactly. I pulled the source. One box is to be singled out and called the starting point. We now further assume that a specific problem is to be given in symbolic form by a finite number of boxes being marked with a stroke. As I read MOS:ELLIPSES, we leave the period in the quotation mark, then do nonbreaking space, ellipsis, space, and pick back up with the text. The alternative would be to gloss the text, which would be something like (condensing) "...and called the starting point. [A] finite number of boxes [are] marked with a stroke." —C.Fred (talk) 02:45, 25 March 2023 (UTC)

Unrelated but I don't enjoy the fact this one HTML comment started an edit war.

Someone, i guess^{(talk i guess|le edit list)}

02:54, 25 March 2023 (UTC)

@SomeoneIguess: So are you saying you agree with the punctuation per the IP's edits and only meant to strike the inline comment? —C.Fred (talk) 02:59, 25 March 2023 (UTC)

Just checked - turns out this is exactly what happened lol. Still didnt warrant an edit war - they could've just redone the punctuation and removed their HTML comment and it could've been prevented

Someone, i guess^{(talk i guess|le edit list)}

03:03, 25 March 2023 (UTC)

The punctuation, in my opinion, seems to be correct seeing this conversation. I'll restore the latest version using ". ..."

Someone, i guess^{(talk i guess|le edit list)}

03:01, 25 March 2023 (UTC)

Algorithm defined as a finite sequence of instructions

I'm not sure if this is a good definition as algorithms may contain loops that sometimes may not stop (unless you close the program) and yet be correct, for instance a service that runs on the background Luiz Felipe de Barros Jordão Costa (talk) 01:40, 24 December 2023 (UTC)

Merge "Algorithmics" into this page

Anyone against merging Algorithmics into Algorithm? That page seems almost completely useless. Weebney (talk) 19:55, 23 January 2024 (UTC)

In favour - go for it. I don't think we need the illustration - not for the term anyway. Amanda Lawrence 10:57, 25 January 2024 (UTC)

It is finished. Weebney (talk) 06:02, 7 February 2024 (UTC)

Contrast with heuristics

While I understand the thought behind contrasting algorithms with heuristics, I think it makes the page longer than it needs to be. It doesn't add new information about algorithms nor does it clear up common misconceptions. It seems like more of a detour in what should be a concise article. Any thoughts on keeping or removing the heuristics paragraph? Clubspike2 (talk) 11:56, 25 June 2024 (UTC)

Definition of heuristic

The statement that "a heuristic is an approach to solving problems that do not have well-defined correct or optimal results" is simply wrong. A heuristic is an approximate and often unproved method, independent of whether there exist well-defined correct of optimal results. See Heuristic in WP: "a technique designed for problem solving more quickly when classic methods are too slow for finding an exact or approximate solution, or when classic methods fail to find any exact solution". Zaslav (talk) 00:16, 19 December 2024 (UTC)