Hacker News Reader

Comment by brudgers | Hacker News Reader

The structure of these projects is a long way from abstractions that deal with "Oh Shit, I'm lost and I'm late" well. The projects are largely inward focused and there's a lot of developers talking among themselves instead. The Open Street Maps page has no "Fix an Error" link. It has Help. Typing in an address and getting irrelevant data isn't a user error. It's a bug. Help is the wrong abstraction.

Gnome Maps is a layer of abstraction on top of that. Not only is there no "Fix an Error" link, there's no clear path for an end user reach the Beginner's page in the OSM Wiki explaining how to edit data.

View on HN

> The Open Street Maps page has no "Fix an Error" link.

If you right click on the map, or click the "add a note" button on the right hand side, you can add a note to the map describing a mistake. You don't need to be logged in

How do I right click on my phone? While a bit tongue in cheek, the assumption that I'm on a computer when using a map is about a decade past the tipping point. Ignoring that, while it's good to know adding notes is a possibility, the channel by which it is conveyed, a Hacker News comment hours later, is exactly why the absence of a "fix an error" button is so frustrating. Nobody outside the OSM community looking at "add a note" would know that it is a good way to record an error.

> How do I right click on my phone?

On osm.org on my Firefox on Android I can long press and get the same popup which allows me to "Add a note". There is also the button on the right hand side for adding a note.

> a Hacker News comment hours later

My apologies, I'm in Europe and timezones, eh?

> Nobody outside the OSM community looking at "add a note" would know that it is a good way to record an error.

Can you suggest a better wording? or icon? Or system?

FTR, when you click on "add a note to the map" you get text saying:

> Spotted a mistake or something missing? Let other mappers know so we can fix it. Move the marker to the correct position and type a note to explain the problem.

Better wording "Fix an Error".

Hear hear. When I'm on a computer I usually go to Google because if I go to OSM and type in a nearby major road, it gives me a list of addresses. And the one I want is probably thirtieth on the list and has a form that is obscure. If you can't search for street names in OSM, it shouln't let me try. Instead of giving me useless results, it should direct me to how I can obtain useful results.

"Commercial Road" - no relevant results in two pages.

"Commercial Road malvern" - no results at all. It turns out in Malvern, Commercial Rd is called Malvern Rd. It seems cruel that OSM requires me to tell it exactly what I'm searching for and to get it right. Giving related information to help the search doesn't help it find the location.

"Sydney Road" - no relevant results.

"Sydney Road brunswick" - finally, the first result is listed as "Sydney Road, Brunswick, City of Moreland, Greater Melbourne, Victoria, 3052, Australia", but it's death by too much information. (a) It's buried in a list of very similar results and I can't quite be sure the one I got is the one I want - this becomes a bigger problem with a minor street. (b) who cares that it's in the city of Moreland? If you're going to list the local administrative district name, why not also the state electoral division? Address are place names, and "City of Moreland" simply isn't a place name. It's the name of the territory controlled by the Moreland City Council, but it's not a placename. Combining (a) and (b), take a look at "Sydney Road, Sparta Place, Brunswick, City of Moreland, Greater Melbourne, Victoria, 3039, Australia". What's that even supposed to mean? I cry.

Google suffers the problem in the opposite way, normalising addresses to the most obscure and ambiguous form they could find, preferring road numbers to names, and using "Melbourne" instead of the suburb name. But at least it finds the place I'm after. OSM is barely useable. Not for want of data, but because it just doesn't care to tell me how to find what I'm looking for. It's exhausting. OSM is useless for practical purposes. I want to use it for a variety of reasons, but I can't.

The default search functionality on OSM ("reverse geocoding" in mapspeak) does, indeed, need a lot of improvement. It's actually a separate project called Nominatim[1]. It's on Github and you can browse the issues and see a lot of similar complaints.

It's a daunting problem, parsing a free text query that may be in any language and trying to best match it with addresses that are standardized a hundred different ways all over the globe. In its current state it isn't good enough for everyday use. It's annoying how often a Google Maps search sends me to the wrong state or even the wrong country. But with Nominatim it's almost a given.

This underscores one of the fundamental features/bugs of the OSM project: It's pretty good at attracting mappers, developers, and 3rd-party projects that use its data and APIs, but it's never really been designed to meet the needs of end users. UX, according to the classic OSM schema, is to be implemented by the various 3rd parties in whatever way suits them best. If someone wants to develop and host a Google Maps replacement using OSM data they're welcome to do it (and they can provide their own reverse geocoding solution if necessary) but that's not an immediate goal of OSM itself, according to the current leadership.

[1] https://wiki.openstreetmap.org/wiki/Nominatim

Sure, but that's a funnel problem, not a problem with the level of learning required to do a basic contribution. Someone with a modest level of commitment will probably make it through the sign up and tutorial.

Apparently there's an issue where you think the beginner walk through should be better advertised outside of the sign up process?

The "funnel problem" manifests itself as a data problem. More importantly, the "funnel problem" is a social problem. A modest level of commitment to learning software is much more modest than a modest level of commitment to fixing software. Learning how to use open source mapping software is based on the immediate practical value of getting from A to B. Learning how to fix open source mapping software requires a different set of values, it requires a commitment to a philosophy/ideology/goal of arguable merit. I'm not arguing for or against it. Only pointing out that taking the moral high ground leads to the idea that signing up and working through a tutorial is a reasonable act by ordinary users when the alternative is Google Maps.

I honestly don't expect it of ordinary users.

I was explicitly pushing back against your depiction of contributing: Learning the complexities of the OSM API's and toolings is even lower. The process is complicated with many moving parts and tailored toward commercial level development not end map users. OSM, like any sophisticated database is complex and requires a substantial degree of technical commitment.

20 minutes for a moderately motivated person is worlds away from what you said.

Of course OSM would benefit greatly from further lowering the bar, but like you say, that's a lot of work.

Reply on news.ycombinator.com