What3Words rescue incident | Hacker News Reader

What3Words rescue incident

twitter.com

Oh dear. Not this drivel again! When will it die?

I would have expected the HN crew to be able to spot the obvious shortcomings of W3W a mile off. Even if much of the supposedly tech savvy media seems to think this idiocy is genius and continues to blow smoke up their arse.

The only impressive thing about W3W is the indefatigability of their relentless PR department.

https://stiobhart.net/2016-01-15-stupidest-idea-ever/

Off by one character.

Is there an free/open software solution? One that is or could end up being a second best, if W3W is primary?

If so, does it suffer the same problem? Of very similar words that are likely to be misheard/misspelt being close enough IRL that it could be mistaken to be real.

Google open-sourced pluscodes, which seems like a decent open alternative: https://maps.google.com/pluscodes/

What3words is a clever bit of research that should never have been spun into a company.

It's just an implementation of a discrete global grid.

I don't think it is live anymore but there used to be a parody called What3fucks, you can probably guess what type of words it used

“My concerns” needs a supporting link, OP.

https://keswickmrt.org.uk/rescue/base-brown-9/

That's the source for the OP story, not the link to “My concerns”

Sorry I thought you meant a direct source to the screenshot.

There's no info here, which isn't surprising given that it's a twitter thread. Is the issue that the GPS was coarse in a remote location, so what3words give them the wrong 3 words? The fidelity of that app is just limited by the underlying signal right?

Based on comments in the thread, it's certainly possible to get homophone or near-homophone w3w codes that go to places many miles apart.

It strikes me the solution is to go back to basics - a numerical GPS location is usually given with a precision (calculated by the receiver), and the two grid coordinates.

To me it seems the first step is to ensure that the accuracy is encoded into any encoded output, so that a recipient can have an idea of the receiver's perceived accuracy, and thus whether to rely on it.

Next seems to be the need to avoid homophones and partial homophones - tree and trees for example. In an idea world there would also be a check word, acting as an error check and erasure code, to help account for typos and mistakes like plurals etc.

I'm not really sure I understand why everyone wants to get behind w3w, but it does seem like they've missed the opportunity to build in even basic fault detection or tolerance. The accuracy could even just be a prefix letter - Alpha for within 10m, Bravo for 100m, X-ray for 1km and Yankee for anything poorer than 1km (which also avoids letter pronunciations being confused hopefully, in the way C, D, E could be on a poor voice channel used by someone unfamiliar with NATO alphabet).

Reply on news.ycombinator.com