Back in the early days of amateur webmapping, many of us got regularly, righteously exercised about the “Multimap patent”.

The patent is described thus:

“A map of the area of a client computer (10) is requested from a map server (11). Information relating to a place of interest is requested from an information server (12) by the client computer (10). The information is superimposed or overlaid on a map image at a position on the map image corresponding to the location of the place of interest on the map. The information (or “overlay”) server (12) may contain details of, for example, hotels, restaurants, shops or the like, associated with the geographical coordinates of each location. The map server (11) contains map data, including coordinate data representing the spatial coordinates of at least one point on the area represented by the map.”

The patent, filed in 1996, was granted several years later in Europe, the States and Australia. (Read the full US patent, or search for EP0845124B at the European Patent Office.)

It’s clearly pretty wide-ranging; the “map server” and “information (or ‘overlay’) server” describes every mashup since the days when the cynics were calling it ‘Red Dot Fever’ (popularly attributed to Schuyler Erle, though I had it in mind that it was Saul Albert) – never mind ‘Google Maps Fatigue’, or, ahem, ‘Why Mashups Suck‘. Almost every webmap could be subject to this patent.

But, fortunately, Multimap never really showed any signs of exercising the patent. Sure, a reference number sat at the bottom of every Multimap page, but that always seemed not so much “we intend our lawyers to get medieval on your ass”, rather “hey, we were here before these Google upstarts, you know”. Indeed, as far back as 2002, Edward MacGillavry – who has a much longer webmapping CV than most of us – could write that Multimap has never “taken any steps to enforce their patents… I am wondering what the actual benefit is of patenting technologies like this”.

Well, today we know.

Coals to Newcastle / patents to Seattle

Multimap was bought by Microsoft in December 2007. At the time, received opinion was pretty much “well done to the plucky little British guys for getting bought”, at a time when Google’s march seemed unstoppable.

We worried that some of Multimap’s bright sparks would be assimilated, of course – but hey, this is the land of Maps 2.0, they won’t find it hard to get jobs. (Visit mapme.at, it’s great.) We fretted that the sole existing OS Landranger slippy map would be taken down, to be replaced by some crap Virtual Earth cartography. That sort of level of furrowed-brow, slightly-bad-things-might-happen concern.

We had forgotten about the Multimap patent.

I am not kidding. I really don’t remember anyone even mentioning it. Searching today, I can see that Rob Dunfey at gisconsultancy.com had been smart enough to spot this angle. Directions Magazine remembered, too, and asked Multimap founder Sean Phelan – grandfather of webmapping, and holder of first the patent, then a very large cheque – about it in passing. He was non-committal. (Probably thinking about his yacht, rather marvellously named Nerdvana.)

Flossie!

Fast-forward to the present.

Earlier today, I idly looked at the latest Google News results for ‘openstreetmap’. I noticed a new site: www.flosm.de, built by an outfit called 123map.

As the name suggests, it’s done in Flash – and as the author of Potlatch, the Flash OpenStreetMap editor, that interested me straight away. It’s a slightly bizarre concept: it superimposes TeleAtlas vector data on OpenStreetMap tiles, so you can compare the two. (There are potential bandwidth-leeching and licensing implications in that, but that’s by the by.) I haven’t seen a great clamour for this, but never mind. Any road up, it’s not going to change the world.

But the story I’d found in Google News was… more interesting. You can read it here (softpedia.com), but here are some of the choice quotes.

“123map revealed that, in order to build Flosm and additional services that it is yet to deliver, it licensed technology from Microsoft, taking advantage of the company’s intellectual property licensing program…

“‘[123map] recognizes, like so many technology companies across Europe, that a way to spur growth in our industry is to license patented technology for mutual benefit,’ explained David Kaefer, general manager of Intellectual Property Licensing at Microsoft…

“It is the specific Microsoft technology that enables this comparison, by making it possible to overlay information on a map image.”

Specific Microsoft technology? flosm.de is built with Flash and appears to call serverside Perl (e.g. http://www.flosm.de/tools/geo03/geocoder_flash.pl?thm=98). I don’t see any Silverlight or .net. I don’t see any Microsoft technology.

Then one of the denizens of the #osm IRC channel pointed me towards the German-language OSM mailing list. Detlev Reiners of 123map has posted there several times, and most recently, has explained a bit more about the Microsoft connection. I don’t speak German, but fortunately Google Translate does. Besides, “Sean Phelan” is the same in any language.

“The patent describes the overlay geographic data, which from different servers on which the data is subject to the client delivered. They are then included on the basis of their geographical Coordinates superimposed. Inventor Sean Phelan.”

In other words, the same Multimap patent we were so worried about seven years ago.

So there you have it. Microsoft is actively claiming a patent for putting stuff on webmaps.

Where now?

Think how, if applied, this changes the webmapping landscape. Every Google Maps view: a penny to Microsoft. Google Earth? Several pennies, I reckon. Yahoo? OpenStreetMap? Anything using OpenLayers? A new startup, like CloudMade or Geofabrik?  Start saving. You owe Microsoft. 

And at a tidy royalty for pretty much every map on the web today, the $50m price for Multimap looks good value. Seen the other way round, the patent looks like a very, very canny investment by Sean Phelan.

Of course, it’s not quite that simple.

Back in 2002/2003, when we were all getting worked up about the patent, Proper GIS Types looked into it and concluded that it was so much baloney. Carl Reed, an OGC chap, found several examples of prior art (1 , 2). A presenter at an Ordnance Survey-sponsored conference in Cambridge the same year concluded both that prior art existed, and that the patent was merely “the automation of a manual process” anyway.

But Microsoft must know this. They’re big enough and – surely? – not so stupid that they wouldn’t try to enforce a patent with enormous negative PR potential… unless they thought it might stand up. And it’s a fairly familiar patent strategy: pick on the little guys first, get a steady income stream. Once the precedent is established, go for the medium-ranking sites. Then, eventually, go to Google and say “I think we should talk”.

I hope Google’s lawyers have their answer ready. I hope it’s not too far removed from Arkell vs Pressdram.

Future Publishing has a spiffy subscription management website, myfavouritemagazines.co.uk (cf this, but clearly not as horribly naff as this).

It does everything you could possibly want, online, which is clearly both convenient for the customer and efficient for the company. And they’re quite proud of it: ”The links below are designed to give you complete control over your subscription.” “This area is designed to answer your every need.” All of that. You can order back issues, get a replacement for an issue that hasn’t arrived, change your delivery address, and so on. It makes me quite ashamed of our simple little online order system at WW.

Except the one thing you can’t do is cancel a subscription. “Please write to us at any time during the initial 60 day period [sorry, the what?] supplying your full name, address, subscription number (if possible) and magazine name. Write to: Future Publishing, Unit 4, Lathkill Street, Market Harborough, Leicestershire, LE16 9EF.”

Free clue: making it difficult doesn’t stop people cancelling their subscription. It just makes them pissed off with you and less likely to buy one of your other magazines.

“Museum of Zoology: The most startling thing about this museum is its levitating Finback Whale, which looms over the surrounding buildings in a strangely out-of-place way.”

I’ve just not renewed the hosting for geowiki.co.uk (well, strictly speaking I’ve not renewed the hosting for geowiki.co.uk every day for the past couple of years, but you know what I mean). So there endeth possibly the first GPS mapping site, sadly neglected since 2004ish, with comedy flat-file storage but a background in a lovely shade of blue. (I’m keeping the domains, of course. ‘geowiki’ is too cool a name to let go.)

“this is an idea that is similar to one i have beenworking on in my mind”

I still sometimes wonder about a site that would document “what’s it like?” rather than “what’s there?”. Probably, in fact, not a classical wiki: every time you replace others’ content with your own, the view becomes a little narrower. (Must. Not. Mention. Wikipedia.) But, you know, hours in the day and all that.

As a general rule, every time I read some Debord I get seized with the urge to do something psychogeographical on the web.

“I just noticed Geowiki, which is about the only way I’d like interactive maps to work on Wikitravel.”

Curiously my GPS tracks are still saved in ~/Geowiki/Tracks/GPX/ . Intermittently I paste them onto a big Illustrator file called GMoE – Geowiki Map of Everywhere. Where “everywhere” means “the UK”. National Grid references are so much easier to work with.

Now, if you’ll excuse me, I’m off to find that Levitating Finback Whale.

I made a flying visit after work yesterday to CloudMade’s product launch in Holborn. Obviously the OpenStreetMap angle interests me personally, but I’m also interested from a waterway mapping perspective. So here are a few disconnected thoughts on both the launch and the products.

Cartography

Three new styles were unveiled, courtesy of Stamen Design: ‘Fresh’ is classic cartography with a understated design sensibility, ‘Pale Dawn’ a mashup-friendly fade-out so you can cram in moar pushpins, and ‘Midnight Commander’ – well, that was the gee-whiz launch of the evening (though it doesn’t appear to be live at time of writing). Will be used for precisely nothing sensible ever, but it’s great to see how a piece of imaginative cartography can seize people’s imaginations. Really encouraged to see these.

CM’s own enhanced ‘Original’ style is a good derivative of classic OSM, with added POI icons making it a good fit in urban areas. I’m a little underwhelmed by the ‘Tourist’ style, but again, it probably works better in urban areas than out here in the countryside, where the choices at each scale don’t really gel.

(Going to London always makes me feel like such a bumpkin: you know, country chap turns up in Barbour jacket and has to buy one of these strange Oyster card things. Where have all these people come from? Why is every inch of space devoted to trying to sell me stuff? Why are the pubs all shit?)

One little detail I appreciated: not all the typefaces are Deja Vu. There’s a very, very slight sprinkling of something Helvetica-like. Little steps, little steps. (Incidentally, why isn’t anyone designing better shields for Mapnik? The current ones are the weakest part of OSM’s cartography.)

Then there’s the Style Editor, where you can take one of these styles and play with it. The user interface is really nicely done and the apply-refresh-apply cycle is tolerably fast, though obviously not as nippy out in the real world as it was at the demo.

It’s not really a serious cartographic tool, and doesn’t try to be: you can’t adjust the line weights or casing colours, for example. But Peter Campbell would have loved it, as it’s essentially a whole webapp devoted to changing the fockin’ colours; and the word that kept cropping up in the presentations was “brand”. In other words: dear MegaCorp, we offer you the chance to get maps on your website with the same palette as the rest of your design guidelines. And since you’re a MegaCorp, you’ll be punting out several zillion tile impressions a day, so may we introduce you to our keenly-priced serving plans? Follow the money.

Anyway, despite that, here’s something I knocked up in 15 minutes. Pretty basic, but it shows how immediate the whole process is. I don’t think I’ll be returning to the Style Editor much in its current incarnation – I’m too much of an Illustrator weenie for that and it doesn’t really give the control I’d want for a custom canal map. But it’s good fun.

One footnote, and this isn’t a criticism aimed at CM in particular, but OSM in general: I think Steve Chilton deserves some more public credit. Most OSM cartography ultimately derives from the osm.xml file which he lovingly tends and keeps in good cartographic order despite the ever-present pleading to add more POIs, more ridiculous tags (smoothness, designated etc.), more Multi-Coloured Swap Shop landuse, and so on. Everyone knows that OSM maps are rendered with Mapnik, but not many know of the essential link between Mapnik and OSM data.

Routing

CM’s map website has also added routing. Now the big push at the event was that “it’s not just routing, it’s a routing API”, and I can see that. People are going to love hacking on this. In particular, I hope Andy adds it very very quickly to OpenCycleMap, because OSM is getting a lot of traction in cycling circles at the moment, and this is going to push people over the edge.

And the reason why: the cycle routing has the makings of being brilliant. It’s the one product from the evening that I can see myself using often. Ask it for a route from A to B, and it’ll find one that uses the NCN where possible, but not religiously; one that follows minor roads in preference to trunk roads; and one which, despite all that, is reasonably direct. 

It works pretty well. If I ask it for a route from Charlbury to Burton, for example, it heads north to Enstone, hugs minor roads parallelling the ‘3400 up to Shipston, then follows NCN5 as far as Redditch – where, very sensibly, it takes a detour on RR55 into southern Birmingham to avoid the schlep via Bromsgrove. Then it’s a mix of NCN, minor roads and canal towpath into Burton. An appropriate and enjoyable route, all in all.

It does like B-roads a little too much: certainly here in the Cotswolds, the B-roads tend to have loonies in 4×4s whizzing along them at silly-mph. Conversely, there was some chat at the event that it shouldn’t be quite so scared of trunk roads; if it’s a choice between 200m on a trunk road and 10 miles on a minor one (or, worse, “no route”), most cyclists will grit their teeth and go for the trunk road.

Edit (Tues 17 Feb): the weighting appears to have been tweaked and it actually looks worse to me. Charlbury to Birmingham, for example, now basically follows the A3400 for most of the way, only diverging significantly after Henley-in-Arden. Perhaps, just as the road directions have “Fastest” and “Shortest” options, there should be “Touring route” and “Commuter route” options for cycling.

The major thing that needs enhancing is the textual directions, which are rudimentary to say the least. They take OSM data a bit too literally, so when a road ‘flares’ on the approach to a roundabout, you get told to take a ’slight left’ or to ‘continue’ for 26m. An example of its over-literal approach:

Now strictly speaking that is, indeed, the way to get out of Charlbury through Enstone, but it’s not very easy to follow. This would be better:

Follow B4022 for 4.6km
In Enstone, cross A44 (slight right/left)
Continue on B4022 for 0.8km
Turn left and follow unclassified road for 0.3km
Cross B4030 (slight left/right)
Follow unclassified road for 23.6km 

And of course, for cycle routing, it should tell you which NCN route you’re on. Maybe this’ll happen when (pretty please) it arrives at OpenCycleMap.

But generally, for a fresh-out-of-beta launch and considering the limitations of the OSM data, the routing is very good indeed. On which subject, Steve asked me to pass on to Fake Steve that he thought taking the piss out of incomplete data was out-of-order. I said I’d pass it on to Fake Steve next time I saw him. So, Fake Steve, consider yourself told off.

Speaking of incomplete data…

Some numbers paraded by the speakers: most of Western Europe is going to be complete by the end of 2010, the UK by the end of 2009, OSM is pushing at 100,000 users already and will have a million by the end of 2010. I think I’ve remembered those right. Oh, and it’s going to be a $100 billion market.

Other stuff

These are the two areas I was most interested in. A quick flick around the rest: some APIs, sadly no Perl, AS3 is a third-party component. Location management will either be massive or astronautical – some hints that there’s much more to come on this. More CM launches at Where 2.0, apparently. Mobile stuff isn’t really my bag as I doubt any of it will work on a £9.50 PAYG phone, but hey.

Apparently CM’s business model is along the lines of: charge for large-scale tile serving; some advertising on maps (with a revenue sharing deal with developers, is that right?) by partnering with existing ad networks. I think. Certainly you can’t fault their ambition – though as someone pointed out, there are lots of Atlantic flights to be paid for. I presume they’re going to need a second round of VC.

Lots of big guys there – DCLG,  Yahoo, Google, Intel (I think) – and comparatively few OSM hackers.

Third-party demos: Where Can I Live? is a very neat idea nicely executed. MapMeat (sorry, mapme.at) takes the smart approach of working with services that you already use, which – with a bit of polish and a few UI tweaks – may be enough to carve out a hefty Twitter-sized niche in a space that others are looking keenly at. Also, when I click “home”, it tells me “Yay, home, don’t forget to feed the cat”, which has got to be good. Apart from the fact the cat is currently in the cattery.

Getting dotted lines in ActionScript seems to be one of those old chestnuts. The ever-resourceful Senocular has a neat ActionScript 2 solution, ported to ActionScript 3 here.

It’s pretty good, but wasn’t quite right for my needs. First of all, it spends quite a lot of time dealing with curves, which I don’t need – polylines are enough. Secondly, it just lets you define one ‘on’ length and one ‘off’ length, so you can do 6px on and 3px off, but not (say) 6 3 1 3. 

So here’s my 2am answer. Pretty compact, seems to work happily. I’ve not genericised it into a DashedLine class or such because that wouldn’t work for my project and I have no reason to think it will for yours, but feel free to do so. I think it’s a bit shorter than Senocular’s, too, though that might just be due to multi-statement lines…

dashedline.as