Talk:What3words

Editorialising in the lead
made a bold edit to the lead, stating in their opinion that this system cannot be used for safety-critical applications. I reverted per this bold edit, per WP:BRD: The WP: ONUS is on Gollem to justify why this material should remain. Repeated reinstatement contrary to WP:brd is edit warring. I suggest you self-revert until you can resolve these issues. --John Maynard Friedman (talk) 22:41, 3 May 2021 (UTC)
 * This opinion is uncited.
 * This statement does not summarise body content, per WP:LEAD


 * I disagree on both points and most other things you write. I didn't state that system cannot be used for safety-critical applications. User 213.205.242.45 did this with a sourced edit . You attenuated the wording, making it too vague . My edit was meant to fix that the statement should reflect the implications stated by the source ("not suitable for safety critical applications") at two points. (1) The use for emergency services should be mentioned as clear example of the need for an "unambiguous location". (2) The ambiguity will instead of may limit the value of the system, since it occasionally will send people to a wrong location in situations when such delays are fatal. I deliberately did not undo your rewording to the version of user 213.205.242.45, but decided to make a smaller change so we could iterate to a wording you can agree on too. Unfortunately, you decided to revert instead of trying to find consensus at that point.
 * Thanks starting the discussion now and for for explaining your problem with the location in the lead section. I do agree with that this does not belong in the lead section and I moved the text to another section. Also, I changed "limits" to "can limit", in the hope this might solve your other objections. Feel free to suggest another solution or explain your objections more clearly.
 * Gollem (talk) 07:29, 4 May 2021 (UTC)
 * However we got there, today's version is a significant improvement, so thank you. BTW, the citation you mention was only to report the security researcher's finding that homophone locations are not as widely separated as claimed: it was WP:synth to declare that this would make the system unusable. (There is an obvious workaround but it would be blatant synth/OR to add it).
 * Taking into account WP: weasel v MOS:CLAIM, we should prefer 'says' to 'claims' - except in the cases such as the performance of an algorithm where the output is measurable.


 * We should not lose sight of the fact that relaying a string of numbers over a scratchy radio link is vastly more error-prone. Warts and all, this system is significantly more valuable to emergency services than grid references or lat/lon in such circumstances. Of course there is the added safety net that a 999/112/911 call over mobile automatically sends the geodata of the caller (but who may have left the incident site to get a signal). --John Maynard Friedman (talk) 08:22, 4 May 2021 (UTC)


 * I'm glad to hear that you find my last edit an improvement. If you want to change a 'claims' into a 'says', go ahead. I don't understand your reference to WP:synth, though. The cited source literally states 'not suitable for safety critical applications' in the title. If it is WP:synth to conclude the found problem 'limits its value for safety-critical applications', it would be because it downplays the implications stated in the source instead of that it exaggerates them.
 * An advantage of 3 random words over 12+ digits of a coordinate pair is that it can be transferred over a faltering mobile phone faster, not that it is less error-prone. In both cases it is needed to repeat and confirm the words/numbers to check for mishearing. It is easier to remember 3 words and it hasn't the problem of dozens of notation variants of coordinates (lat-lon, lon-lat, decimal degrees, degrees-minutes-seconds, etc.). A disadvantage to me is the fixed accuracy of 3 metres, while with coordinates I can define a location with any accuracy, e.g. decimal degrees with 2 digits (1 km) to 8 digits (1 mm). But the main problem is that one needs a device to encode a location to three words and the other side needs a device to decode it to coordinates. If both sides have such devices, why not use it to send the coordinates as digital text in stead of reading out words? Gollem (talk) 10:22, 4 May 2021 (UTC)
 * I suspect that W3W for emergency services will fade away because Advanced Mobile Location will replace it with direct data exchange as you suggest. I am surprised that it is not already in the See Also, I'll do it now. --John Maynard Friedman (talk) 16:19, 4 May 2021 (UTC)
 * I see you alphabetised the see also list of alternatives. The list was in chronological order and used to give the year of introduction of each system too. I liked that . Should we reintroduce the years? Gollem (talk) 16:44, 4 May 2021 (UTC)
 * I didn't spot that sequence; most see alsos are in alphabetic order. I don't really have a preference so go ahead. But see next before you do, maybe? --John Maynard Friedman (talk) 18:14, 4 May 2021 (UTC)

Proposed new section: alternative systems
Before I do a lot of work on it, can I check that there are no serious objections to a new section on alteratives? This would cover:
 * Grid references: in places like Great Britain (OSGB) and Ireland (OSI, OSNI) where maps are usually printed showing the national grid, it is possible for someone with a map but no other technology than a scale ruler to identify a location correct to 100 metres
 * World Geodetic System, aka WGS84, a world wide system. This grid is printed on some European maps, possibly US too?
 * Or do I mean Universal Transverse Mercator coordinate system (UTM) which seems to have come to the fore.
 * Mapcode favoured by Google Maps: just excerpt the lead.
 * Advanced Mobile Location again, excerpt the lead with maybe a bit more emphasis on emergency service.

Thoughts? Does it stray too far off-topic? --John Maynard Friedman (talk) 18:14, 4 May 2021 (UTC)
 * The source would need to relate the other system to W3W, and that is how it would need to be presented here, otherwise it would be off-topic. Alexbrn (talk) 18:34, 4 May 2021 (UTC)
 * I disagree with Alexbrn. There is no need to make this restriction. But I would like to limit the list of alternatives to more or less the current topics under "See also". I would definitely not include national grids (there are literally hundreds of these: more than one for most countries). If a specific international geographic coordinate system (lat., lon.) is mentioned, it should be the scientific ITRS supported by the UN and not WGS84 (which is used only for a GNSS operated by the USA). Gollem (talk) 19:07, 4 May 2021 (UTC)
 * This is the article about W3W - if RS compare it to something else, that's fine, but we can't add things just because we think they are similar. We already link to Geocode where readers can find more general information. SmartSE (talk) 19:14, 4 May 2021 (UTC)
 * Yes we can. If we keep the text concise, we can list the obviously related systems of the topics under "See also" with a few explanatory words per topic. Gollem (talk) 19:27, 4 May 2021 (UTC)
 * To avoid WP:OR, which is prohibited by policy, there would need to be a source making that connection. If it's obvious, such a source should be easy to find. Encyclopedia articles reflect what good sources are saying about topics, not what editors think might be neat. Alexbrn (talk) 19:29, 4 May 2021 (UTC)
 * You can't stretch the restriction on WP:OR that far, otherwise list articles would not be allowed either. W3W is for geolocating, so we can make a section on a small selection of other methods for geolocating. (We do need a source that it is a geolocation method for each method). Gollem (talk) 19:46, 4 May 2021 (UTC)
 * I agree with Alexbrn that it would be OR to include them without sources comparing them to W3W. SmartSE (talk) 20:02, 4 May 2021 (UTC)
 * I have to say that "original research" was the last challenge I expected. "Other methods exist to achieve the same result" is surely WP:the sky is blue? I foresaw WP: fork (though many articles contain summaries of other articles) or even WP:synth (here is this and there is that: is it possible to compare and contrast them without synthesis?) But I'm sorry, I still can't see from anything written above that demonstrates original research. --John Maynard Friedman (talk) 08:06, 5 May 2021 (UTC)

Our disagreement might seem larger than it actually is, as we could have completely different ideas on what this section would look like. I suggest to keep it really simple, to stay far from WP:OR. For example something like only grouping of the topics of "See also":

See also

Geocode systems based on geographic coordinates in chronological order:
 * Universal Transverse Mercator coordinate system (c. 1942)
 * Maidenhead Locator System (1980)
 * Geohash (2008)
 * MapCode (2008)
 * Open Location Code (2014)

Emergency location sharing systems: Gollem (talk) 13:43, 5 May 2021 (UTC)
 * International Cospas-Sarsat Programme (1979)
 * Advanced Mobile Location (2014)


 * That is pretty much what I had in mind, though I thought maybe we could attach at least its wp:short description to each, to give visitors at least some clue as to why they might want to actually see also. --John Maynard Friedman (talk) 11:16, 5 May 2021 (UTC)

Reads like an advert for WhatThreeWords
Most the of the material parts of this Wikipedia page reads like promotion rather than a factual article. Even supposedly factual statements are dripping with a sense of impartiality. If the page is promotional then it should clearly state that. — Preceding unsigned comment added by Matt Peacock (talk • contribs) 17:38, 18 September 2022 (UTC)


 * read Matt Peacock (talk) 17:39, 18 September 2022 (UTC)

Anonymous User appears to be sanitising this article of any possible criticism (eg removing discussion and links to articles where emergency services have criticised W3W and replacing with content where they have praised). Moreover, their edits are poorly described (eg edit described as moving content into a different section is actually an edit which removes a material amount of content, which happens to be negative about What3words). What is stated here is a summary, and the user’s full edit history should be reviewed so that one can form their own view. Matt Peacock (talk) 18:54, 19 September 2022 (UTC)


 * Agreed - looks to be a continuation of COI editing from someone connected to W3W. I will take a look and revert where necessary. SmartSE (talk) 09:44, 20 September 2022 (UTC)
 * The reason I initially came along to edit was seeing a link to a tweet of mine that grossly misrepresented what I said, by the same anonymous user. Cybergibbons (talk) 11:21, 20 September 2022 (UTC)


 * Now coming from but per very similar edits in July:   they are almost certainly the same person as 81.109. SmartSE (talk) 11:53, 20 September 2022 (UTC)
 * It's worth pointingout that the IP corresponds to what3words.direct.quickconnect.to - a Synology NAS drive someone has named what3words. Now, anyone could do that, but it's a strong indicator something is amiss here. Cybergibbons (talk) 13:18, 20 September 2022 (UTC)
 * For a guess, a reputation management agency at work. . --𝕁𝕄𝔽 (talk) 15:44, 20 September 2022 (UTC)
 * It was almost certianly What3words themselves. Please be vigilant about material that uncritically promotes w3w, or links to puff pieces doing the same Ms7821 (talk) 20:30, 2 February 2023 (UTC)

Tierney citation
For the moment I think we need to retain this citation: because it makes important challenges to the company's assertions (as reported by RSs). But it fails WP:PRIMARY and WP:BLOGS (because the author is not personally notable). It would be better if we could replace it with a report in RS that confirms it. Can anyone supply? 𝕁𝕄𝔽 (talk) 16:16, 20 September 2022 (UTC)


 * Awkward one that - what does RS mean? It was printed in Mountain Rescue Magazine, and there has been limited secondary reporting. Cybergibbons (talk) 17:57, 20 September 2022 (UTC)


 * That would mean Reliable_sources -- i.e. your blog is not a peer-reviewed academic journal.


 * I know some academics knowledgeable of DGGS, and I can tell you that the interest in doing or confirming research about w3w is zero - partly because the system is already deemed useless, and partly because (let me quote the article) «The company has pursued an assertive policy of issuing copyright claims against individuals and organisations that have [...] reverse-engineered code that replicates the service's functionality». The chilling effect is very real (and after seeing this week's events, I'm going to say it's even sought). --  Iván Sánchez (talk)  19:03, 20 September 2022 (UTC)


 * RS = reliable source. That's not to say that your blog is unreliable, but we try to base content on sources where there is some editorial oversight. Thanks for the info about Mountain Rescue Magazine - I found it online here (page 30 of the pdf). Another possibility would be to cite This from Techdirt which summarises the blog. SmartSE (talk) 19:05, 20 September 2022 (UTC)
 * Mountain Rescue would qualify as a wp:RS (peer-reviewed academic journals are only required in exceptional cases) because it is a serious publication with editorial control. Unfortunately they've published it as an OpEd without analysis or comment, so it is not unambiguously better. But better certainly, in the sense that if they thought it a load of rubbish they would not have given it a two-page spread. I suggest we use it for now, with Andrews's blog going into External Links. Does anyone disagree? (Smartse, I've corrected your page number, I assume you don't mind? Also TechDirt is a blog too, so no good). --𝕁𝕄𝔽 (talk) 10:11, 21 September 2022 (UTC) revised 10:13, 21 September 2022 (UTC)
 * Thanks. It was collaboratively edited with them, but that still doesn't stop it being OpEd. Cybergibbons (talk) 18:04, 27 September 2022 (UTC)
 * The key factor in making it usable is knowing that they exercised some editorial judgement and control. --𝕁𝕄𝔽 (talk) 00:04, 28 September 2022 (UTC)

Criticisms section
I see from the history there used to be one, but now the criticism is incorporated into the Design and Reception parts. Any objection to recreating the section and moving some stuff around? WikiNukalito (talk) 09:20, 22 April 2024 (UTC)
 * Generally we have felt it better to embed criticisms in the relevant sections rather than hive them off to the end of the article where they have no context (and less likely to be even read by readers on mobile [aka, most]). Also, IMO, doing so would weaken sections like 'Design and Reception' because it would leave them stating the company line without balancing response. Can you be more specific about what you have in mind?
 * Meanwhile, the lead as it stands says nothing about any criticisms. WP:LEAD says that it should summarise the key points of the body and this is certainly key. Would you rectify that issue in any case? --𝕁𝕄𝔽 (talk) 10:56, 22 April 2024 (UTC)
 * That sounds perfectly reasonable. My only thought was that when looking at it with all the sections collapsed (default on mobile I think) it looks like there are no criticisms, but a sentence in the leading paragraph would also account for that. Something like
 * What3words has been subject to a number of criticisms both for its closed source code and the high potential for ambiguity and confusion in its three word addresses . This has resulted in some to advise against the use of What3words in safety critical applications.
 * Where the refs are to existing sources (Eden comments, Arthur paper, Tierney blog, Mountain rescue article) WikiNukalito (talk) 15:57, 22 April 2024 (UTC)
 * That reads fine except for the word "high", which would need explicit and neutral sourcing. Maybe "risk" might be a better word than "potential"? The citations need to be formatted properly, do you need help to do that? 𝕁𝕄𝔽 (talk) 17:03, 22 April 2024 (UTC)
 * It is the criticism itself which has claimed high potential e.g. the entire Tierney blog or from the abstract of the paper "What3Words has also attracted criticism for being less reliable than claimed, in particular that the chance of confusing one address with another is high. This paper investigates these claims and shows that the What3Words algorithm for assigning addresses to grid boxes creates many pairs of confusable addresses, some of which are quite close together. " which I think supports the statement, though if you have a better way to word it I'm fine with that.
 * Since the page already references those sources I was just using the  syntax (with the slash the right way !) WikiNukalito (talk) 17:20, 22 April 2024 (UTC)
 * Per WP:synth, we can't combine sources to reach a broader conclusion. But I think "significant risk" is just about ok. "High risk" or "high potential" are a step too far. Leave it to readers to draw their own conclusions from the sources.
 * Yes, the named references are needed for multiple use but by formatting I meant using cite journal, cite report or just cite web. 𝕁𝕄𝔽 (talk) 17:45, 22 April 2024 (UTC)
 * I don't think that is synthesising, the quote above says "What3Words has also attracted criticism for ... the chance of confusing one address with another is high", which is how I read the first sentence. As in, the reason people are criticising it is because of the high risk. Nonetheless, I think significant is fine, but I was saying high because that's what the source says! WikiNukalito (talk) 18:12, 22 April 2024 (UTC)
 * I've made the suggested edit (and fixed the refs), I think it's supported by the text of the sources cited, but if you decide to change 'high' to 'significant' or similar it I won't complain. Cheers. WikiNukalito (talk) 05:21, 23 April 2024 (UTC)
 * I don't think that is synthesising, the quote above says "What3Words has also attracted criticism for ... the chance of confusing one address with another is high", which is how I read the first sentence. As in, the reason people are criticising it is because of the high risk. Nonetheless, I think significant is fine, but I was saying high because that's what the source says! WikiNukalito (talk) 18:12, 22 April 2024 (UTC)
 * I've made the suggested edit (and fixed the refs), I think it's supported by the text of the sources cited, but if you decide to change 'high' to 'significant' or similar it I won't complain. Cheers. WikiNukalito (talk) 05:21, 23 April 2024 (UTC)