North Atlantic Radio episode 2

North Atlantic Radio

Brian and Matt talk about APIs from Yahoo and Google

In this, our second episode, we’re talking all about maps and geolocation APIs. We test the new Yahoo Placemarker API live on air and end up very impressed. We also chat about the first impressions of the new Google Maps Data API. For future reference, the maximum length of the data you can send to Placemarker is 50,000 bytes.

Brian’s books

Matt’s books

As you’ll be able to tell, we’re still tinkering with the audio settings. Bear with us in these first few podcasts as we work out how best to record it all.


Brian: Welcome back, everybody. This is our second North Atlantic Radio podcast. It’s a special one because this week I’m actually in Glasgow along with Matt.

Matt: Yeah, we stretched the budget of North Atlantic Radio to pay for Brian to fly over.

Brian: Yeah, only economy.

Matt: You’ve been on bit of a trip.

Brian: Yeah. I was travelling around, at a conference last week and then popped in and visit some friends in London and then came up to Scotland and staying in Glasgow for a few days and just generally catching up on lots of reading. I’m on holiday. So they put me to work.

Matt: Yeah. And make you do podcast stuff. Yeah, so we’re both in Glasgow. Nice evening today — we should actually be outside because it’s a gorgeous evening outside.

Brian: That’s the nice thing. We can just tell the listeners. This is a podcast from the balcony.

Matt: Yeah, overlooking … well, some good bits of Glasgow and some bad bits. Good view, though. We can see Ben Lomond apparently, I always get told that. What are we gonna be talking about this week, Brian?

Brian: I think you’ve somewhat alluded to it a little bit with Ben Lomond: we’re gonna be talking talking about maps this week because it’s been quite an interesting topic. Strangely, we we chose this several days ago, and we been kind of researching in a few things and it just exploded in the last two days with the Where 2.0 conference and some big announcements. So we’re going to talk I think a little bit about maps and just kind of go through the standard stuff, our standard format.

Matt: And, the same as we did last week, finish on what we’ve been reading this week.

Brian: So this week was the Where 2.0 conference, which is all about where you are and how the Web and mapping data — just geo location, have really taken off in the last year or two. But I think this week both Google and Yahoo announced, uh, some API interestingness. Do you want to talk about those?

Matt: Yeah, well, we’ve done a bit of research in these two, and we want to talk about one because one is good and will probably skip the other one because it looks quite boring. The good one is Yahoo Placemaker API, which allows you to pass in arbitrary strings or even URLs to pages, and it will try and yank out geo data from that string. So if you’ve got a blog post about a trip you took to — well, Brian might write a blog post about his trip to Belgium and Glasgow and then back to Reykjavík. And if you passed the blog post in, Yahoo Placemaker ideally would pull back the cities of Glasgow, Reykjavik, London and the countries. So you’d be able to do interesting things with that, such as place blog posts on a map.

Brian: A couple of years ago there was a company — I think it was Gutenkarte — and they did some really interesting work where they had taken older texts — things like the Odyssey and the Iliad, as well as more modern ones like, uh, War of the Worlds — and made a pass of the text. It then starts plotting the story on a map. So you kind of actually get more context where you understand why these two little Greek city states were always fighting. In text you don’t understand that they might be right next to each other, or there was a tiny little island in between them that they were always fighting over. Once you start putting it onto a map, it adds another dimension of sort of understanding.

Matt: And we’re going to try a live test of this.

Brian: Yeah, we’ve chosen two things. First, we’ve chosen the North Atlantic Radio about page, because it already says where were normally calling in from. So do you want to run that? See what we get?

Matt: Yeah, well, on the about page — I mean, you can go to it while you’re listening, if you’re in front of a computer. We mentioned the fact that Brian’s in Reykjavik in Iceland, the fact that I live in Scotland, that I moved from Edinburgh to Glasgow. So there’s a few places in there, so we shall run it now and see what the Yahoo API comes back with?

Right. So I can see the screen and Brian can’t at the moment. It’s come back as XML, and I can tell you … not sure what this first bit is, but there’s a “super name” of northern Europe, so we can maybe we can work that out later. But it’s got Edinburgh, Scotland, with a confidence of seven. That’s good. Glasgow, Scotland with a confidence of seven. Reykjavik in Iceland, confidence of seven. Scotland the country, confidence of seven.

Brian: Yeah, I think it’s on a scale from 1 to 10

Matt: And [laughs] and Atlantic, New Jersey, with the confidence of four.

Brian: Fair enough, fair enough. If you set your threshold at something like a six or higher.

Matt: Yeah, so you could say if it’s above six out of 10 confidence you include it, otherwise you ignore it. Uh, the final place it mentions as well is Edinburgh University, which, OK, yeah, I mention that in my biog, so that’s quite good. It’s not just necessarily —

Brian: Country names

Matt: — and places of interest as well. So I’d say that did a pretty good job. Simple data but —

Brian. What was the confidence level on the University of Edinburgh?

Matt: Uh, seven — Oh, it’s sorry, I should keep scrolling down. It’s got Reykjavík again. Oh, OK, yeah, and then it references all the — it’s quite good, actually, if you get the XML, you can choose two formats to have your response returned in. One’s XML, one’s RSS, and we’re looking at the XML response. And the first half of the data is the places. So Edinburgh and then the longitude and latitude of the point. And then the second half is the references it found. So it gives you the fact that it found Reykjavik and it gives you an XPath expression to the point in the document where you find that reference, so you could double-check things.

Brian: Yeah, [unintelligible]. Basically, you can — not reverse engineer it but you know with confidence where it’s pulling these things from. And this is interesting for me, because it’s also microformats-aware. So if you’ve added microformats to your page and you’ve said explicitly, this is an address this is a country, it seems to be pulling it out and giving it a higher confidence level as well. This is another excellent example of bigger services, you know, understanding and being aware of any semantic information you put in.

Matt: Yeah, I’m very impressed with Yahoo’s geo APIs, they’re always a lot of fun. I’ve not used them for anything other than tinkering with, but it’s always great fun, even if you just go and look at old blog posts and see, you know, how much you can pick out. And you could just have a play around and see if you can plot everything you’ve written on a map, and it’s good fun. And if anyone’s got any ideas of how you could use that Yahoo API, um, use it to do something more interesting.

Brian: Or maybe you already have. Just let us know.

Matt: Maybe you have, maybe you’re ahead of the game.

Brian: What’s our email?

Matt: hello at, which has nothing to do with New Jersey.

Brian: So if you if you have done something, just let us know and maybe we’ll talk about the next week, or at least put a link to it.

So the other thing we’ve decided is our bio page is pretty obvious, So we’ve thought of a more difficult one to see how it fares. And one of the famous examples that I always remember is the book White Fang by Jack London, and we’re going to see how well it might pull out things or how many false positives. It’s just so we’re aware of the limitations and expectations of the service.

Matt: Yeah, So we’re looking at we’re going to run the Placemaker API. Well, we’re going to pass it the URL to the Wikipedia page for Jack London and see when it comes back with.

Uh, first of all, errors. Uh, this is the trouble with live demos. We actually didn’t test this beforehand. The Wikipedia page for Jack London exceeds the maximum length of the data you can pass through. [Laughs.] So we can’t do that second part of the test.

Brian: Well, maybe we’ll come up with another URL

Matt: We’ll run it afterwards with something to do with Jack London, and put the results in the show notes so we can say how well we thought it did.

Brian: Because he had — what was the, what was the daughter’s name? Virginia?

Matt: There was some, there was some dodginess involved. Christ, he had … someone had an abortion. Someone refused. Someone turned the baby over to an ex-slave. And the ex-slave’s name was Virginia Prentiss. We thought it might pull out Virginia the state, and we want to see what confidence level it had.

Brian: So you’ve got Jack London, London as a possible city, Virginia, as a possible place and name. So, yeah, it’s been an interesting one to see what happens.

Matt: We’ll play around with it and if we can’t get it [working] then you can always have a go and send us the results.

So the other, uh, map based, uh, announcement at Where 2.0 this week — or today, in fact, as we record this, was the Google Maps Data API. Neither Brian nor I have played with this, um, and it certainly doesn’t seem to be getting as much press as the Yahoo API, partly because this possibly could be quite boring. It could be one of those ones where, if you delve into deeply, it’s actually something fantastic. At the moment, it just seems like it’s cloud based storage for KML files or KML data.

Brian: Most likely, you can read, write, search through, and upload geo-related data so that it could be data points or some sort of shape and then be stored on Google’s end. And so it can be queried from all the different properties and mobile. And it’s like that. Hopefully then might give people an opportunity to mash up some information a lot easier, because it’s just all in one place, but certainly not as fun as the new Placemaker.

Matt: Yeah, Simon Willison mentioned the Google Maps Data API on his blog this morning. Um, I guess I think he expected it — as as I did, and probably most people did — expected it to be, um, like a replacement for a geospatial database system like PostGIS or something like that, and so he had a play around with it this morning and he said, it’s actually quite boring and didn’t do anything as interesting as he’d hoped. If Simon Wilson thinks it’s boring, it’s probably not worth having a look at.

Matt: Finally, a little skit we do at the end of these podcasts. We talk about what we’ve been reading since the last episode.

Brian: I think I’ve powered through several books since the last time.

Matt: You’ve got a library to take back home with you.

Brian: I love coming to the UK, I can load up on inexpensive books from Amazon. But I think last time I was reading — I don’t even know the last time … we plan on doing these things every three weeks, so it’ll be interesting to see how many books I can get through, especially when I’m not on vacation. But I read through Fooled by Randomness, which is an interesting book. It’s the precursor to Black Swan. I finished the other day How to Lie with Statistics, which is a really interesting book, because once you understand how you can lie with statistics you can really spot it all over the place ans you can really call people out on it. I’m currently working on two books: Information Dashboard Design — which is good for a few projects I’m working on — and Calendrical Calculations, the third edition, which is incredible. It’s all about all sorts of different types of calendaring systems and how you can map Mayan calendars to the post-French Revolution metric.

Matt: Does it have the code to do the conversions?

Brian: The book is just chock full of hardcore math functions on how to run them and, you know, go between the two as well. It’s it’s fascinating all the little hiccups because you think the biggest calendar problem was Y2K, but the book goes on and on about how piss-poor Excel is, and actually just figuring out something as simple as, uh, leap years and stuff. So it’s really fascinating.

Matt: Well you found a fantastic little nugget about NASA, where it sounds like their space shuttles just store the day of the year as an integer. So January the first is day one. January the second is day two, and it just goes on: 3, 4, 5, 6, 7. But then, once it gets to December 31st, it doesn’t roll over the next day, back to one again. It just counts it as 366, 367, 368.

Brian: They purposely skipped space shuttle missions that roll over year end because they don’t actually know what will happen.

Matt: Yeah, it was 1993 or something?

Brian: No, it was 2006.

Matt: 2006, where they delayed a shuttle launch from — December the sixth, was it? — till January the next year because they knew that the date would be out of sync on the shuttle with the —

Brian: — with the ground service —

Matt: — yeah, which is fantastic.

Brian: NASA of all people.

Matt: At least it’s good to know that they have, they’re aware of it, and they’ve got to work around. But the fact that some guy in NASA just thought, how do I store dates? Screw it: an integer.

Brian: Exactly, because I mean, it makes sense. How long is a space mission? It’s never going to be more than a year old anyway. You don’t need to worry about carrying out 365 days. It makes me now wonder about the International Space Station [laughter]. Some guys like pressing, “Please send food”, and, like, nah, day 700. Sorry.

What have you been working on?

Matt: What have I been reading? Not much, actually. Well, Brian just pointed out a book to me. Thanks. Brian. You’re like my librarian.I just moved through from Edinburgh — what, six weeks ago? Something like that. I can’t really remember. Um and so I’ve just been, rather than reading any novels, I’ve been reading books about Glasgow to try and get used to the city. But one of the books I picked up from the city library is called Public Sculpture of Glasgow, which is a book for nerds. How many pages do you reckon it is? 400? 500? 540 pages. Just about sculpture on buildings in Glasgow. So I go for a wander around the streets, look up, see something I like, go back and look at the book.

Brian: You don’t carry it around the streets with you, do you?

Matt: No, it’s way too big. That’s that’s what I’ve been reading.

Brian: Well, thanks for everyone that’s given us feedback so far. We’ve gotten quite a lot. We’re going to tweak this version of the podcast for some of the suggestions that you’ve made, but we’re always open to more. You can certainly email us at hello at, or just grab either one of us on IM or any of the other services when you see us online. We’re looking forward to, you know, continuing on this and talk about interesting things. Thanks a lot.

Matt: Cheers guys.