I know it would be better to raise this with Mapbox directly. I already did, on 2017-12-26. So far I haven’t received a response.
Last year (OK, last month. OK, 3 weeks ago) I decided to play with Leaflet. Using OSM tiles, everything went as expected. Decided to give Mapbox tiles a try in Leaflet, got a Mapbox key and went ahead.
The Mapbox tiles are around 3-4 months out of date. I can see things I added/changed 3-4 months ago but nothing more recent. Did a forced-refresh with ctrl-F5 in the hope it would trigger a re-render and waited a couple days. No change. So then I contacted them via a form on their website and still haven’t had a response.
Today I had a thought. Mapbox are trying to shift people to their vector renderer, so maybe their tileset renderer isn’t getting the love and attention it used to get. I went to one of the demos on their website, painfully navigated half-way around the planet (on my old computer Mapbox vector rendering is a lot slower than OSM mapnik rendering) and zoomed in on my locality. Still 3-4 months out of date.
So, is anyone else seeing this? Is it something I’m doing wrong? Have Mapbox forgotten about the UK? Are Mapbox ignoring Cardigan? Did I do something really stupid in a changeset 3-4 months ago that breaks the Mapbox renderer and stops it processing more recent changes? Or are other people seeing this?
As far as I know, Mapbox is no longer directly importing new OSM data into its dataset, they’d rather verify new data before that. I don’t know if it’s the same time for all regions, but there are edits I made in Moldova in early October that are not yet visible in Mapbox.
I don’t know why they’re doing that, OSM quality checking is nothing that should be done outside of the project. However, to my knowledge, Mapbox has always done some data refining on their own.
I wonder how they’re going to verify the data. I appear to be the only person actively mapping in my area, and before I started mapping about 6 months ago most of the edits were no more recent than 3 years ago. Mapbox certainly didn’t contact me about the edits of mine I made more than 4 months ago, so I have no idea who they’re verifying them with.
Looks like they’re just slow. Very slow.
I don’t have any problem with them adding stuff by themselves or changing stuff by themselves. After all, any mapper can do those things. But holding data back seems silly.
The reason I investigated is that of late I’ve had to contact various small businesses to attempt to clarify details (usually postcodes or other aspects of their address). I figured that one day, one of them might ask me how to incorporate OSM mapping onto a web page. Leaflet looked like a good way to do it, with Mapbox tiles if the business was big enough to generate a lot of hits (few around here are). But there’s no way I can recommend Leaflet/Mapbox to anybody. “Yeah, I’ve just added your details to the map, so if you use Leaflet/Mapbox to add a slippy map to your website your business might show up in three months, if you’re lucky”. Not that anybody has asked yet, but I like to be prepared.
So can anyone recommend an alternative to Mapbox? I like the Mapbox terms: free for low usage, pay if your usage goes up. If anyone around here got enough hits for Mapbox to require payments it would probably have resulted in a big enough increase in business to justify the payment. But with a 3-month lag before changes appear? Forget it.
If anyone from Mapbox is reading this, let me make this perfectly clear. If anybody asks me how to add a slippy map to a web page I don’t (yet) know what solution I’d recommend but I would strongly recommend they avoid Mapbox.
There’s a list of some options at https://switch2osm.org/providers/ , though not all of those providers fit into the same niche (and Mapzen, which is on that list, will cease to be at the end of the month).
The easiest answer there is probably “add a leaflet-based map to your existing website” and serve either your own tiles or someone else’s (it’s unlikely that a simple “how to find us” page for a small business would come close to free usage limits for any provider that has a free tier).
I’m also very unhappy about this. What also makes me feel bad about this, is that as a customer you can request to have edits incorporated more quickly. So areas with few commercial users will see only sporadic updates on the future, while others are updated more often?
So yeah, they really have a “data team that reviews all edits”. Sounds like a very bad idea to have this separated from normal OSM. What happens if they discard changes that remain on OSM? Will they fork OSM? What will happen to conflicting changes? Sounds like one of their worst decisions ever.
There’s a few alternatives such as Carto or MapCat, but none of them is as sophisticated as Mapbox
Even the largest of the businesses around here (other than the county council and jobcentre) rely on shared hosting. They don’t have the expertise to set up their own tileserver even if their hosting plan permitted it (which it almost certainly doesn’t).
I’ve had a quick browse at your suggestions and I’ve yet to find one with a free tier (other than Mapbox). The places I’m thinking of are probably paying around £100/year for hosting (typical prices from hosting companies local to here). And given what they’re actually doing with their sites they could get a hosting package that meets their needs for around £10/year (a bit unreliable with crappy speeds but more than adequate for their needs). One of the cheaper tileserver tiers I found was about £370/year.
So if anybody asks me I’ll just have to tell them to use Google for their slippy map. It might not have as much detail as OSM but at least it’s not as slow to be updated as Mapbox when people report changes.
I just had an idea. Mapbox could offer both “raw” (up-to-date) data and “verified” data (3 months old), then make “verified” data a premium service and charge more than for “raw” data. Except that in most cases people would prefer the raw data in the first place.
It’s not that difficult - there’s nothing in https://switch2osm.org/manually-building-a-tile-server-16-04-2-lts/ that wouldn’t be run-of-the-mill to someone who’s looking after an existing Apache server, for example. If someone’s just paying for a little hosted managed Wordpress somewhere then yes, that won’t support it, but for anyone who has an actual website with existing JS on it then adding a Leaflet-based map really isn’t a big deal.
That sort of price isn’t a million miles away from what you can get a virtual server for for serving map tiles as well as an actual website (it’s a little less than I’m paying, but I’m rendering significantly more of a map both in terms of detail and coverage than any local business would want to just to have a “how to find us” map).
Er, OK - clearly you know best and there are no alternatives out there
Ah, it all makes sense now. They’re slowing things down to milk more money out of customers. You don’t just have to pay for the traffic (fair enough) you have to pay to get your changes on the map quickly.
If they’re doing it for a better reason then I hope they’re aware that a lot of people are going to interpret what they’re doing as a money-making exercise.
If it wasn’t to make more money, it’s a damned stupid idea. If it is to make more money then they probably realize it’s a damned stupid idea but desperately need the money anyway, which doesn’t look good for their continuing existence.
I wasn’t worried about sophistication. OSM rendering would be adequate. I can think of ways the user interface could be improved for the benefit of end-users (I know it’s not aimed at end-users) but it’s still good enough. The only reason to go elsewhere for tiles is to avoid the strain on the OSM servers. I doubt any business around here is big enough to cause a problem, but Mapbox’s free tier was the obvious way to go just to be safe.
I just had a quick look at MapCat. Nice (visually pleasing) rendering. A few days out of date (which is nothing compared to Mapbox). It appears you must pay to be able to use it with Leaflet. And you have to sign up to marketing spam to find out how much you’re going to pay. Oh, and big problem (at least with their ordinary map): it doesn’t show house names. Not even when the house only has a name and doesn’t have a number (which is very common around here).
Carto looks like I have to sign up for marketing spam even to see a demo of what their stuff can do. There’s an old saying “If you have to ask the price, you can’t afford it.” My version of that is “If I have to ask the price, you don’t want to sell it.”
Nope, it’s not that difficult if you know what you’re doing. If you’re in the business of running web servers or hosting web sites, it;'s not that big a deal. However, even if you’re in that line of business there’s some time involved in getting it all going, time that you have to hope will be amortized somehow amongst all the people clamouring to use your new tileserver.
The people I’m thinking of have fairly limited hosting that they paid somebody else to put a website on to. The sort of people who ring up their hosting company to complain that two users can’t collect mail then find out the reason for that is that they deleted those users the day before (when they actually intended to delete two different users). That level of technical expertise.
Indeed. But you’re supplying the OSM/tileserver expertise yourself. There are a few businesses around here that have attempted to build their own website (on shared hosting) and the result looks like something that would make a GeoCities user feel ill.
You’re almost as pedantic as I am. I’ll be a little more precise. So far I have yet to find any OSM-derived solution that meets all my criteria, therefore the best solution I can recommend, should anybody ask, is to go with Google. I’m not happy with that, but at the moment that appears to be the best option.