Rich data for every location on Earth, with Ed Freyfogle from OpenCage

[00:00:00] Mike: Hello, hello, and welcome back to APIs You Won't Hate. I am your co-host today, Mike Bifulco, one of the, uh, I guess founders of APIs you won't hate, and I'm hanging out with, uh, my good pal, Phil Sturgeon. Phil, good morning. How are you today?

[00:00:14] Phil S: Hello. I should know where I am today. Usually you ask me where I am and I'm like, today I'm in Bristol and it's lovely and I'm having a great time.

[00:00:20] How are you doing? There we go. I'm doing

[00:00:22] Mike: great, man. It's, uh, it's good to see you. Definitely like one of these nice proper spring mornings over here while we're recording. Of course, you know, when it releases, it could be, uh, anything weather-wise and, and wherever you are. But yeah, having a nice morning over here.

[00:00:34] Um, and I'm, I'm happy today. We're gonna get to, uh, sit and meet and chat with our new friend here, uh, ed Fry Fogel, uh, ed. Thanks so much for joining us. Um, appreciate you being here. You are working on a project called Open Cage, among other things. Um, tell us a little bit about yourself and, uh, how

[00:00:49] Ed Freyfogle: you got to where you're now.

[00:00:50] Hi guys. Thanks for having me on the show. I am a longtime listener, uh, longtime API practitioner. Um, yeah, so the main project I work on is a company called [00:01:00] Open Cage. We have been doing that now for about eight years. Uh, it's a geocoding api, so we offer forward and reverse geocoding. For those that don't know forward geocoding is you have an address or, or let's say, uh, some sort of text string that refers to a location and you wanna know the geographic coordinates of that.

[00:01:20] And reverse is the opposite. You have, you have geographic coordinates and you wanna know the, the location description. Uh, and that's what we do. Uh, that's it. That, that sounds like a little bit, actually, it keeps us quite busy. Um, we. I guess the differentiator of our services that we do this only with open data, so we can talk a bit about exactly what open data is.

[00:01:42] But, uh, listeners may have heard of projects like Open Street Map is probably one of the most famous, um, open data sources. We we're, we're heavily involved in the open street map, but also there are other open data projects as well. Um, yeah, so, so happy to go deep on all of that and talk about kind of the challenges of.[00:02:00]

[00:02:00] Of running an API based business. Particularly what, I mean our biggest competitor is Google Maps. You might have heard of them. It's a Right, yeah, sure. Small company out of California. Yeah. Yeah. So, um, uh,

[00:02:11] Mike: I'm, I'm kind of curious to hear the genesis of Open Cage. So where did the need come from and how long ago did you start building it?

[00:02:17] Ed Freyfogle: Well, we, we, I, I originally had a, another business. I was co-founder of a business many years ago. This is, this is really gonna date me, but, um, Around 2005, 2006, Google Maps came out and for the first time this, you had the amazing phenomenon that you could put points on a map, right? And you could move the map around.

[00:02:39] I don't, I dunno if you remember this. I mean, kids today assume this was all how it always worked, but that wasn't the case. Uh, and so this was a huge innovation. And so we, um, like many other people had the idea that we would put, do, use this in, in the real estate space. And so we built a, a real estate search engine, kind of a, a metasearch engine where we would aggregate all the [00:03:00] different real estate listings from different, uh, sites and started the company do that, ran that company for about 10 years based in London.

[00:03:07] But we served 10 different markets, um, and uh, you know, did quite well. Uh, anyway, one of the key challenges there is, of course you have to geocode. So we typically would get the, the property listings as addresses. And to put them on the map, you need the longit latitude. And so we would start, you know, figuring out ways to do that, using external APIs, trying to use our own, uh, software.

[00:03:30] Um, and it was challenging. I mean, some countries, some countries it's very easy. Uh, for example, in the UK you, if you have very precise postcodes, uh, some countries have very good address data. So we were also serving markets like Brazil and India. The much less good data up there. Um, I should also mention that while this was going on, those, the 10 years that around the company was also coincided with the exact, the founding of Open Street Map in London, which, which initially started in 2004.

[00:03:57] Um, and then kind of really took off. And so [00:04:00] we were very active kind of in that scene going, going to some of the, the pub meetups and things like that. So got to know this space better and better. Um, anyway, fast forward to 2015. We eventually sold this real estate business to one of our competitors. Um, but in the years leading up to that, in like from like 2013 or so, some of our customers, our customers were big media companies in these different markets.

[00:04:25] And some of them had said, how, how do you guys do the geocoding? How do you, you know? And so, um, we said, well, we have this internal technology and, and basically they asked us if they could use it. And so we kind of thought, let's spin this off as a, as its own project. And we did that under the name Open Cage.

[00:04:40] Um, but it was just really still very much in fledgling form when the acquisition happened. So it was very much a beta product. Uh, so the buyer was not interested in that. And so myself and, and one of my colleagues at the old business, he, he and I came the co-founders of Open Cage and we, we took it over when the real estate business was sold [00:05:00] and we've been running it ever since.

[00:05:02] So, so it's been going now since 2015. Slow but steady, uh, growth, you know, that there's, there is a lot of demand for geocoding, and we can, we can talk a bit about that, about why that is. But, um, yeah, and, and we've been, it's, it's, it's our, it's a, you know, it, it pays the rent. So that's what we,

[00:05:21] Mike: yeah, sure. No doubt.

[00:05:22] I think it's really interesting to be a serial co-founder and your, your second, um, company is essentially the brainchild of the first company, uh, kind of like a. A bit that broke off and, and became its own success story. That's super interesting. Um, have you found that along the way, um, the things you learned building your first company, the real estate search engine sort of helped to inform what you're doing now?

[00:05:45] Well,

[00:05:46] Ed Freyfogle: the two, the two companies were quite different in that, and when they, and they said this is quite common, right? For the second business, you wanted to do the exact y you're so fed up with all the problems of the first business. You wanna do the exact opposite. And, and that was kind of the case with us as well [00:06:00] in that.

[00:06:00] The first business, first of all, it was a much more of a traditional startup in that we, you know, we had external funding. We eventually had about 20 employees. Um, it was kind of more your traditional startup. Uh, but our customers were mainly big media companies. So the people who run the big, uh, real estate websites of the world, which on the one hand was good because they were, they would pass a lot, but, but it was, you know, very one-off bespoke kind of deals, long negotiations, that kinda thing.

[00:06:30] With, with open Cage. Our business, particularly in the beginning, it was much more self-service. You just come to the website and stick your credit card in and you can start geocoding. Um, much less kind of customer interaction. Um, very tight. I mean, one, the, one of the reasons we liked being an API business is much less emphasis on, um, User interface.

[00:06:53] I mean, of course the API needs has a good interface, but you don't need to spend, you don't need to have a, spend all your time designing and things like that. [00:07:00] Um, much less SEO susceptible. Our first business was very dependent on seo, which when it's great, it's great, and when it's bad, it's very bad. Um, so we kind of went exactly in the opposite direction.

[00:07:13] Of that. Um, and I, and, and it's a, there's no funding, there's no external funding. It's, it's just purely funded from profits and from our, in the beginning, from our own savings and things, um, from selling the first business. But, um, so kind of the opposite in many regards. Uh, sure. So, yeah, this,

[00:07:33] Mike: by, by now, this may feel like an obvious thing to you and maybe to the people listening to this, but, uh, I'm, I'm curious what your, has your value proposition changed as, um, The years have gone from, hey, this is an alternative to Google Maps, to, hey, this is, uh, you know, maybe the first place you should turn, or something like that.

[00:07:50] Why, why were people looking at alternatives to Google Maps to begin with? Well,

[00:07:52] Ed Freyfogle: as, so really we have, we have kind of three competitors, so, so one competitor is definitely Google Maps. Google Maps is the giant in the [00:08:00] space. There are a few other big, um, proprietary players. There's a company called here.

[00:08:05] Basically they do kind of car navigation services and typically their model is that they. They collect their own data. You know, Google has their cars driving around and things like that. Um, and it, it, it's, the data is theirs. They, they, and they run the whole system. And then if you want to use their data, you have to of course pay them in some cases quite a lot.

[00:08:26] But also you have to agree to their terms and conditions, which can be quite onerous. Um, particularly around attribution, about how long you can store the data. Um, can you keep storing the data after you, if you stop being a customer, for example. Um, you know, many, many players make you delete the data if you ever stop, which, um, but a big issue, of course is cost.

[00:08:47] Um, so Google charges quite a lot. Um, which, which they need to, to justify cuz they have all these teams collecting data, things like that. The other,

[00:08:56] Phil S: and that's not just, that's not just like fender lock-in, that's fender [00:09:00] blackmail. It's like if you try and leave us, you have to delete Correct. Your database.

[00:09:04] Correct. Correct. That's not,

[00:09:05] Ed Freyfogle: that's not cool to, to their credit. Uh, I mean the, technically the product is of course quite good. Um, and, and because of their dominant position in, in the consumer mindset, uh, you know, the database does get updated. I mean, the first thing someone does if they open a new business is update their listing on Google Maps, cuz otherwise they won't be found.

[00:09:24] So, um, so what I always tell people is like, look, if, if you want, you know, if, if you feel Google's the best and you wanna pay for that, then then go for that, you know? Uh, but, but you're gonna pay a lot and you're gonna, if you want good enough at a radically more affordable price, we are your choice. Okay?

[00:09:45] Um, and we're able to do that because we use the open data, but that does. Open a new, um, competitor and so, and, and, and that, that competitor's companies doing it themselves in-house. So, so we [00:10:00] often, that's kind of what we're competing with. People are like, well, I need a lot of view coding. Let me do it myself.

[00:10:05] And that's a difficult argument to have with people. Obviously engineers like to build it themselves. They like to have their own trace to train, set to play with. Those people typically become our best customers once they realize it's a freaking nightmare to run it yourself. Um, or also when, when, you know, maybe the, the guy who ran it, um, leaves the company or whatever, you know, and then ev you know, some new person has to maintain it and they're like, actually this is complex.

[00:10:31] 1, 1, 1 point to make. There that in one way, though, I guess we're different probably in many of the APIs that you've covered here on the show or that people think about, is. And we, there really are two pieces to our system. One is, of course, the software, like any other api, but secondly, it's the data. The data is changing all the time.

[00:10:51] And because the world is changing all the time, so the O Open Street map gets about four to 6 million edits per day. [00:11:00] Um, so, so this is kind of a living creature, uh, and that always needs to stay updated. So the there is, and, and the volume of data, of course, is massive. So this is kind of a DevOps challenge, if, you know, for, for, to stay on top of that.

[00:11:13] And it's not something where you, you know, it's not, you just, you install it and then you forget it and, um, you know, maybe once every six months you upgrade to a new version or something. It's not like that. Um, so, uh, so that's why many, many people then turn to experts like ourselves to, to handle that

[00:11:30] Mike: for them.

[00:11:31] Of course. Yeah. So behind the scenes then I get, I would imagine this is me just, uh, wager a guess. That you're probably not living in the world of, um, a AWS lamb does and things like that. Are you running like a mess of servers in data centers all over the

[00:11:44] Ed Freyfogle: place? Yeah, we have our own servers. We don't, we don't use, um, AWS or any other cloud provider.

[00:11:51] We we're very happy with Hener. I don't know if you're familiar with Hener in Germany. Uh, not, not just in Germany. They have other locations as well, but, but they're very good. Um, but [00:12:00] yeah, we have our own dedicated server set, multiple locations. Of course. Um, Just cost wise, it wouldn't make sense. I mean, with that volume of data, yeah, there's no way you can do it cost effectively.

[00:12:10] And this is actually one of the arguments I make when I, when people are like, oh, we'll do it ourselves. I'm like, if you do the math, it's gonna be very difficult for you to do it cheaper than letting us do it for you. Certainly, yeah.

[00:12:22] Mike: Yeah. So, so when you started building the product, when Open Cage was born, um, I'd imagine there was probably some basic use cases in mind, uh, right.

[00:12:31] And, you know, beyond maybe just forward and reverse Geocoding, was there, like, were there certain languages or, um, I don't know, was there, um, uh, a Venn diagram of customers you were after versus customers you were not after for the,

[00:12:43] Ed Freyfogle: uh, api? Well, it's very interesting because there are many, many different types of geocoding and many different use cases, and.

[00:12:50] One of the challenges that we face is people come to our service and they're very kind of fixated on their, what, what they need. And [00:13:00] they think, you know, that's the only way to geocode. So many, many people think only of forward geocoding. They don't think of reverse geocoding, which reverse geocoding is actually massive because it's um, kind of the vehicle tracking space.

[00:13:12] Um, basically the cost of the cost of a tracking device and the accuracy of a tracking device and the performance of a tracking device In terms of. I've gotten so much better. Right? So, um, more and more things have tracking devices, you know, obviously vehicles, but you know, bikes and things. So, um, so, you know, so there's more and more data being, being general there.

[00:13:38] But the other point is then thinking about, you know, of course if you send us some coordinates or you send us an address, we can return the opposite. But what we also do now is, um, Kind of crosslink this with other data sets. So there are many, many different ways to refer to locations and many different types of statistical codes, you know, uh, uh, ranging from the obvious ones like iso [00:14:00] country codes to, but like all these things, like in the eu, there's a thing called nuts codes, which is a way that people refer to locations or, um, in the US you have FIPs codes, um, all kinds of other relevant information.

[00:14:12] For example, you might ask, you know, what time zone is this location in? And, and so we have tons of different, uh, data sets that we kind of cross reference and thus make it simpler for developers. So that usually geocoding is only the very, very first piece or, or a tiny piece in a data processing chain.

[00:14:32] Okay? So in some sort of ETL system or whatever. And so the next step then is, okay, now I got the coordinates. Now I need to feed that into this other data set. So, so we're trying to do things like that for the developer so that they don't need to, you know, take the time to do that. So, um, right.

[00:14:49] Mike: Yeah. I was, uh, playing around in your, um, demo for the Geocoding api, uh, earlier and was kind of struck by the amount of information that comes back, like.

[00:14:57] For, for folks listening to this, I think you should [00:15:00] challenge yourself to think of what would happen, what you would expect back from an a p I if you typed in your home address, for example, or your office address, uh, and asked it to return geo code information to you. To be honest with you, a few things came to mind for me, which are probably obvious latitude, longitude, country, and things like that.

[00:15:15] But the, the shock that I had was that there's a breadth of information that, uh, open cage returns that are like really thoughtful things that, uh, obviously make sense, but once you see them, they make sense, right? Like, uh, time for sunrise and sunset, what flag emoji to use for that particular location, uh, what country it's in.

[00:15:32] Things like regions and things like that, that are just like, there's so much information about a given, you know, dot on a map. That I can see this, um, fulfilling lots and lots of use cases. And I can imagine that, uh, as time has gone on, you've added lots of features here that help a lot of people get things done without having to go from one API to the next, to the next to chain this

[00:15:51] Ed Freyfogle: altogether.

[00:15:51] Yeah, we, we've expanded it quite a bit over the years. Um, but yeah, it's the most obvious example. Mike. Uh, uh, you're in the US right? So if someone types in a US [00:16:00] address, uh, US addresses do not contain the county. But they very often, you might wanna know which county I'm in. Uh, so like this gets into some of the use cases that we didn't quite anticipate.

[00:16:09] So for example, um, payment providers. So let's imagine if a transaction happens in a certain county, maybe then they need to assess a different tax rate, for example. Um, so things like that, uh, where a, you know, humans have developed almost an infinite number of ways to divide the world up and categorize it.

[00:16:30] And, and, and. Um, and talk about it and assign codes to it and whatever, and, and with tons of weird historical anomalies, you know, and, and so that's that chaos, you know, that, that, that's kind of what we try to simplify and present a clean, obvious interface too. Um, I mean, Phil, you're in the uk, which in my opinion, having spent 10 years living in the uk, uh, is.

[00:16:55] Possibly the most insane country on earth in terms of [00:17:00] admin hierarchies. And I, I mean it started, it starts even, yeah. It starts even with the fact pretty, which country are you in? Right? I mean cause Cause you know, you could say, I'm in the UK and they'd be like, no, I'm in England. No or no, I'm in like as well as a country.

[00:17:12] Is it not a country? Exactly. We're

[00:17:14] Phil S: a country of countries and that right off the bat is fucking ridiculous. So, So, um, yeah, absolutely. There's, I, I used to work, I've, I've worked with, um, geocoding and Reverse Geocoding a bunch of different times for the last couple of jobs. I mean, uh, all the rage doing all the solo mo stuff in New York.

[00:17:30] Um, and, uh, yeah, it was working for a, a carpooling company where we had these really weird scenarios. Like most people just think, I would like to type in the address that I'm going to for work, and then we'll see if anyone's going along with me. Should be easy. Some of these ridiculously rich corporate offices were like, there was one, there was one road where the road was one the word, so it was like one Pepsi drive or whatever it was.

[00:17:54] And so people were typing in the number one Pepsi drive, which doesn't exist because it's actually a, A [00:18:00] one one. And so the building was number one word, one Pepsi drive, and there's all these weird things. So I, I have noticed that at least, um, at least your product will kind of take a guess at a few different levels of, of likely result.

[00:18:13] And so if you don't match on the first attempt, it will kind of try and show you some other things like I just put in my address. And one of the suggestions was, did you mean Bristol? It owned my house, but it also knows the city in different types of place. So it gives you different things you can play with for different confidence levels, I'm guessing.

[00:18:27] Uh, yeah, and

[00:18:28] Ed Freyfogle: I mean, there's a lot going on. I mean, first of all, duplicate place names, typos, um, I mean, in the uk many, many people, they don't use the number zero. They say like, oh, my phone number is oh seven six. They use, oh, so then when they type just subconsciously, They type O instead of zero. Right? And, and I mean, even though intuitively they pray, no, it's a zero.

[00:18:50] So all kinds of things like this about the way people think about numbers and things. And, and this is before we even get into the absolute madness of postcodes. [00:19:00] Um, uh, and then, and then this is before, you know, at the UK to its credit. I mean, you know, many countries are, are at least have postcodes. Some countries don't have postcodes, some countries don't have addresses.

[00:19:14] Very common. Uh, uh, you know, most of the world does not have an address. Um, they use kind of just landmarks and things. So it, I mean, this is a, from a technical standpoint, this is a project you can work on, you know, easily for the rest of my life. It can work on geo kitting. There's, there's no shortage of, of weird cases.

[00:19:31] And, and, and the big issue is a lot of the data that people have that they want to have geocoded is just gibberish. You know, it's bad and um, you know, so we have, it's a lot of working on data cleansing, frankly, so

[00:19:47] Phil S: That's awesome. I mean, what, one question I've got for you. I was going back through, I was working on the, uh, survivor and other people's APIs book, um, self Plugging on the podcast cuz Might as well.

[00:19:57] And one of the things I remember was how, how many [00:20:00] different. Geocoding services have shut down. Like every time I find one that I like that's not Google, it ends up vanishing. And I found the list of ones that was, um, uh, or is it, uh, Yahoo Place Finder. That was really great. Um, Algolia Places vanished Simple.

[00:20:13] Geo Cor bought out by Urban Airship and then just closed down and rolled into their marketing platform. Um, what's keeping yours going when other one seemed to be

[00:20:21] Ed Freyfogle: vanishing? Well, first of all, Phil, I'm not gonna apologize. We're gonna put 'em all out of business one after another. So it was you. Um, you know, you can tell Google we're gonna inform, um, no.

[00:20:33] Well, first of all, I will say I do think we have a bit of an advantage of not having investors, right? I think many of the business, some of the business that you've mentioned, you know, have taken lots of funding. And so they're then under a time pressure to kind of provide a return on investment on that funding.

[00:20:52] And it's difficult. It's difficult. I mean, one of the challenges with geocoding. You know, I can, I can sit here and tell you lots of funny stories about [00:21:00] geography and weird use cases or whatever, and if you do not need geocoding, you are not gonna buy geocoding for me, right? It doesn't matter whether you like me, you like my website, like it is very difficult to induce demand for geocoding.

[00:21:13] So, uh, what I have to do is kind of be present in the market and wait until you have a geocoding project. And then hopefully at that point, you're aware of me, you have a good opinion of me, and, and then you come and test out my service and hopefully we do a good job and then you become a customer. But, um, yeah, but the point is, it, it's, it's very difficult.

[00:21:31] We've been fortunate that we could go slow, so to speak, and we weren't under pressure to, to very quickly, um, provide, provide a return for any kind of investors. Another point is, uh, I mean, Algolia for example, they. I think you're talking about our places search, which shut down a while back. Um, so, so again, the, the, within the realm of forward geocoding, they're kind of different things.

[00:21:58] There's what we call [00:22:00] geocoding, which is you have a complete address and you wanna know the location. But I think what you're thinking, talking about more is kind of, people have like a restaurant name or they want kind of an auto suggest on a page or whatever. And this is really challenging. The challenging part there is not the technology, the priority part is having the database.

[00:22:18] Having a database of all the restaurants worldwide is almost impossible. And, and how do you keep that up to date? And, and this is where someone like Google shines, of course, because they have the consumer demand, because they have, you know, billion people walking around with Android phones. So they're collecting data off of, it's gonna be difficult for a startup to compete in that space in terms of how do you get this data, um, and keep it up to date.

[00:22:42] I mean, now with Covid. You know, we saw this huge, uh, uh, dying off of many businesses. So instantly everyone's database was very out of date. Right. And, and, and you know, of course we try to, to, we rely on the, the open data community, particularly open street map [00:23:00] and, and we do our best to give back and support that community.

[00:23:02] But, you know, there are some areas where open street map is excellent. There are others where, certainly with transient things like restaurants and stores that come and go very quickly or. It might be seasonal. It's a challenge. I mean, there are even, there are even businesses, um, that just help big companies kind of keep their location data up to date and get it out.

[00:23:25] It's, I, I, I'm friends with the guy, he's the founder of a company in London called Loco, and all they do is help these mega organizations keep all their data up to date on all the different platforms. Cuz you've got. You've got Google, you've got Facebook places, you've got TripAdvisor, you've got all of it.

[00:23:40] And, and you know, someone like Starbucks for example, I don't know how many tar Starbucks are. I mean, they're like worldwide. Probably three or four Starbucks every day. Opening and closing. Yeah. That's

[00:23:50] Phil S: impossible to manage. Entitled

[00:23:51] Ed Freyfogle: by. So someone's gotta get that data cuz someone's gotta push it out in the right format.

[00:23:56] Uh, ideally, of course, not just like a, a longitudal latitude, but [00:24:00] like a nice pitcher, the opening hours, all these kinds of things. Um, it's a lot. It's a

[00:24:05] Phil S: lot. Yeah, for sure. I mean, I remember the, um, a while ago, like 10 years ago, while Simple geo was still a thing, somebody very proudly announcing that, like they'd invented this, this brand new kind of geocoding service that was basically simple geo, but, but bonk.

[00:24:20] Which if for our international audience is like one, not particularly major city in the uk. And it's like, yeah, I've, I've got this, uh, service this's basically a data set of every single restaurant in Bourmouth that I bought off someone. And so my, my business is that I put that on a map.

[00:24:35] Ed Freyfogle: Yeah. I mean, no disrespect to the great people of Bourmouth, but, uh, the, there are players who take that approach, uh, in, in the terms of they focus only on a certain country and they try to be really the best at just that country.

[00:24:47] And, and you know, that can work and obviously there isn't a big advantage to having that local knowledge. And, and, but, you know, many people writing software are trying to build platforms that can be used globally [00:25:00] or, or in multiple markets. So for them it's probably better to, it can be a challenge to work with all these different small individual players in the different geographies.

[00:25:08] So, That's also one of the advantages that we try to provide is that we aggregate these different open data sets. So some countries are quite progressive in releasing the governmental data and things like that. Mm-hmm.

[00:25:21] Phil S: Yeah. You've gotta pay quite a lot for UK data, right? Like the Royal Mail has loads of amazing data, but they keep that under locking key, or is Open Street Map replacing that or supplementing that in some way?

[00:25:32] Yeah.

[00:25:32] Ed Freyfogle: The, the, the situation, the UK has gotten a bit better. There is, um, open data from the ordinance survey. Again, there are restrictions. You know, it's not, it's not always, you know, I think there's a delay. There's a six month delay until it's published or whatever. Um, yeah, again, the, the, the situation in the UK has its own level of insanity.

[00:25:50] So, um, but, but nevertheless, the, the, the tide is definitely moving in the direction of more openness and more sharing of data and, [00:26:00] um, and so we try to ride that, that tide, ride that wave and aggregate it all and make it simple for people to use. It's encouraging

[00:26:07] Mike: to see that products like this can be so well supported and, and obviously so well used.

[00:26:11] Um, one of the things that really stands out on your site as you kind of browse as a potential user of, uh, open Cage is that, uh, you support maybe the largest list of, uh, client library SDKs that I've ever seen. Uh, I, and on your site, I think it says something like 30 plus languages and it everything you could imagine laying and go and JavaScript and PHP and all these things.

[00:26:33] Um, that's, that's pretty impressive. H how does that come about? How do you end up supporting all these different languages?

[00:26:39] Ed Freyfogle: Well, we built it up over time. Um, some of them are, obviously when we started we, you know, we had the, the languages that we use internally and, and, uh, some of the more popular languages.

[00:26:49] Um, but basically we are our standing offer to anyone. So, so to you, dear listener, is if you can write an SDK for our api. [00:27:00] Which, which actually our API is really not that complex. You know, we have one endpoint and, and a few optional parameters as a REST api. Um, anyway, if you write an SDK in a language that we don't yet have or we will gladly pay you or, or an integration with, um, you know, different CMS systems or what any type of software.

[00:27:20] And get in touch and, and as long as you open source it, we will list it. We will pay you. We will, we will do a blog post where we give you full credit. We don't, we don't try to claim any credit for it. Um, we want obviously as many SDKs and things as possible, so, and a lot of people have taken us up on that over the years.

[00:27:38] Um, and, and yeah, it's worked out well. Yeah. Yeah. There's quite a few, and, and some of those, some of those are, are actually quite, Active projects. I mean, you know, we obviously, it's not possible for us to know every single language, so it it, we rely on the community for some of these other languages and, but, but it, it's worked out.

[00:27:56] I mean, people are using them. That being said, I do think, as I said, [00:28:00] our, our api, we, we make an effort to keep it as simple as possible, have very few changes. Um, sure. And, and so in that regard, it's, it's, it's a more stable API than perhaps, you know, others that, that have been

[00:28:15] Phil S: featured here on your show.

[00:28:16] Yeah. This is really cool. I was just looking at the list of providers and I noticed that, um, geo Code p h p has built a driver for it. So that's William Duran is, um, that package is a big deal in the PHP community and I've really liked that they have built a provider for it. So it's kind of a generic interface with different, um, plugins on the backend and.

[00:28:34] I was just looking at it. Go, man. Uh, geo Code PHP is getting 300 downloads a month. No, it's not. Your, your provider is getting 300,000 downloads a month. So just that plugin. That's massive. Well, yeah,

[00:28:47] Ed Freyfogle: so, so there are many of these kind of aggregator libraries in, in almost every language has one like that.

[00:28:53] Um, it's a double-edged sword. It, it, what I would say is that, uh, on the one hand, obviously there are a lot of [00:29:00] people who. Don't, they want geocoding, but they don't wanna pay for geocoding, right? So they, they use a library like that and they just go through them all. But, um, the bigger problem is they usually, when you use an aggregator library like that, you're getting kind of the least common denominator of all this.

[00:29:18] It has to provide a kind of standard interface. So, so all these different useful data annotations that we provide typically get lost. Like the, the, the, it may not provide a way to access those and interface with those, so, Obviously we try to promote our own SDKs, which we have more control over, and where we can really make sure it's, it's fitting our, our api.

[00:29:38] But you know, if that's how p we we're not dogmatic and I try to force anyone, you know, anyone who wants to use our api, they're, they're welcome to it whichever way they want. Um, and, and we'll do our best to document that and present that. But, um, Yeah, there, there is a lot of demand for geocoding. The most typical use cases, someone just has a [00:30:00] database of addresses, they wanna stick on a out, and that may range from, you know, 50 addresses to, you know, 50,000 addresses.

[00:30:07] One of the big challenges we face is it is difficult to convert those people into customers. Um, so they, they

[00:30:14] Mike: probably don't want to pay on an ongoing basis or don't think they should pay at all to, you know, put a pin in a map somewhere at that point.

[00:30:20] Ed Freyfogle: I mean, one, one. I, I've made this joke before, uh, but, um, we offer a pretrial so we understand people need to test our service.

[00:30:30] And so you can sign up for, uh, an API key. It takes about one minute, and then that api, you're on a free trial, and with that you can do 2,500 API requests a day, which I think is more than enough for people to test. One challenge we face is somehow there's some sort of thing that happens to software developers with their vision when they read our documentation, is that they go, they go blind to the word trial.

[00:30:54] Mm. They think we're offering a free service. Yeah. Yeah. Okay. They, they, they, they don't [00:31:00] see that. And, and we do not offer a free service. We offer a free trial. So, and then, you know, so if someone's using my service every single day, after a couple weeks, obviously, you know, we notice that we have scripts that catch it.

[00:31:11] And I read to 'em like, look, you're obviously depending on my service, so you know, if you're gonna depend on my service, I I need to be compensated in some way. Of course. Yeah.

[00:31:21] Phil S: I think I've been, I've been guilty of that a bit in the past, like the amount of time I've spent doing like really creative cashing and all these other like really cheeky things to keep the rate limiting down on services.

[00:31:29] When I. Probably would've been better off just paying for it. Like, I've definitely done that. It's just something about the developer mindset that's like, we will not throw 20 bucks a month at it. We will throw one dev, you know, a hundred dev hours at it a month instead. That's much more value for money.

[00:31:44] Ed Freyfogle: Yes, yes. So it's, it is frustrating. I mean, you know, some people, you know, I, I polite Lee contact them, say, Hey look, you've been, you know, every day for the month you've hit the free trial limit. Maybe it's time to customer and say, oh, thanks, you know, and then, but [00:32:00] obviously some people upgrade and some people stop or whatever.

[00:32:02] Those people don't bother me as much as we, there are the people who then, you know, try to sign up a hundred times and things like that, and it's so. I, it's so annoying because it's like, guys, we've been doing this for eight years. Do you think you're the first person we've seen who would like, thought to put a plus on their email address and, and to trick, you know, like, come on guys.

[00:32:21] Well, I like that you have

[00:32:22] Phil S: monitoring system that's like capable of spotting all this

[00:32:25] Ed Freyfogle: good observability. We have a lot of different system to try to. Nudge people to become customers. Uh, you know, so it that, that is, I have to say, one of the most frustrating parts about the business is that we provide, in my opinion, a, uh, a, a good service at a very reasonable price.

[00:32:42] You know, and we well documented and, and we do quite a lot to give back to the community, but the, but the open data community and the open source community and sponsor projects and sponsor events and everything. And then you have people who are just like, can't be bothered to, you know, pay us. As you say, $20 to, to help on that project.

[00:32:59] [00:33:00] It's that, that does burn you out after a while. Well, this

[00:33:02] Mike: is a problem too that has existed in Mapmaking, at least for as long as maps have been a thing. Right. And I feel like the, the grand tradition of Mapmaking is to include paper towns in your, uh, your print maps. Right. Which is like a fake location on a map so that if I, uh, someone copies my map and they see this town that doesn't exist on their map, I know they copied specifically from me.

[00:33:21] Right. So maybe the open cage answer is to just start filling, uh, SDK responses with fake, uh, data after a while. And, you know, slowly, well,

[00:33:29] Ed Freyfogle: well, that, that what we do do eventually, I mean, after many warnings and things, we eventually just start returning random results to people. Oh yeah, sure. There you go.

[00:33:38] And, and we, we, I mean, we tell them ahead of time, we're like, look, dude, I've asked you politely, you know, five or six times to become a customer, but like, if you. You know, at this point I can't help you anymore.

[00:33:49] Phil S: It's better than my suggestion. I was gonna just start returning 1 23 butt street if everything is, but something,

[00:33:54] Ed Freyfogle: no, we do, we do actually true random results so that they probably won't even notice it until [00:34:00] much later when they actually need to rely on that data.

[00:34:02] Oops. And, and that's, you know, revenge is a dish vest cold. Brilliant. Yeah. That, that's one challenge of running a freemium service. Sure. I, I will say that, that we do, it does get a bit frustrating at times, but, but by and large, many people, um, you know, they see the value in what we're providing and, and become happy, satisfied customers and, uh, and it's great.

[00:34:26] Yeah. So let's

[00:34:27] Mike: talk about what else you're working on, ed. Uh, we talked a little bit, uh, before the show about geo mob. Do you wanna tell us a little bit about that?

[00:34:33] Ed Freyfogle: Yeah, so, um, you know, we, we, we. Mentioned briefly there, there's a whole world of interesting things going on in the geospatial space. Uh, you know, weird, uh, weird stories and weird anomalies and, um, but also, uh, a whole lot of innovation.

[00:34:47] I mean, there's a massive amount of innovation, and so for the last 15 years now, I've run an event in London. It started as, it was kind of a meetup of geospatial developers. And it's called Geo [00:35:00] Mob. And we, we meet up once a quarter or so in the evening, a couple people give talks about their projects and then we go to the pub and people can hang out and, you know, have a raging debate about whether England is a country or a sub country or whatever.

[00:35:15] Um, anyway, so that's, that's proven quite popular and, and there's a big community around it. And so now we've also expanded to a few other cities in Europe, um, which is great. And then when the pandemic hit, and we obviously couldn't have our events anymore, we started the podcast as well. So, and usually on the podcast we interview different people who have spoken at the events, who talk about their different projects.

[00:35:39] Um, and it's really interesting. I mean, as I said, there's, there's, there's just been a wave of innovation unleashed by. I would say a couple things. One, uh, first of all, um, Opus Street Map, we're making data readily available to everyone. Secondly, smartphones, you know, now everyone knows exactly where they are all the time.

[00:35:58] And, and, and you have a, [00:36:00] not just, you know where you are, but you have a, a computer that can do things with that information. And so tons of cool services are being built and tons of interesting, um, things about that. That. And then of course, so we just have all these weird, wonderful, wacky anomalies of the geospatial world.

[00:36:15] So, uh, people give talks about all this and, and so anyone who is interested should please come along to one of the events or, or listen to the weekly podcast we're on. We're coming up on episode 200 now, so, uh, wow. Congratulations. Yeah. There's a lot going on. Yeah.

[00:36:31] Mike: So Phil is, uh, one particular, uh, interesting user of, of geocaching.

[00:36:35] I feel like, uh, Or not Geocaching. Geocoding. I feel like at some point, uh, Phil, you can, uh, enlighten us all on your, your Geo journey. Uh, somewhere along the way you might, might be a good guest for the show there.

[00:36:45] Phil S: Yeah, yeah, yeah. I mean, at the moment I'm. At the moment, I'm mostly just putting like trees on a map.

[00:36:51] I've just got, you know, we're planting a hot thousands of trees. I mean, God, we're doing 8,000 in a week soon. So we're doing a lot of tree planting. [00:37:00] Um, and we, for many of those projects have to take a photograph of where the tree is and get that on a map. And some people have told me that I should put all the trees on open street map cuz there are layers for that sort of thing.

[00:37:12] Ed Freyfogle: There are people who do that. Yeah. There are people who tag individual trees, of course, with the, you know, the scientific genus and all that. And, and, um, yeah,

[00:37:21] Phil S: we can do it. Um, yeah, the, um, Trouble with ours is that is kind of like a proprietary data set in that like people pay us for those, for that tree data.

[00:37:31] So if I then shove it onto open street map, then they, they, those people didn't need to pay me to plant those trees. Am I lose all my funding. So at some point when, when the woodland is a bit grown and like we can kind of, you know, tag the mature trees and, and then the woodland becomes quite lovely, that that's something I'll probably do, but.

[00:37:47] There's all sorts of other weird stuff we've gotta start doing. Like we're looking at, um, various, um, open weather API to see how much rainfall they're getting so we can kind of preempt which ones are gonna struggle and go water them. And looking at what soil types [00:38:00] there are so that we can try and plant, um, plant clay soil earlier in the season, um, before Christmas, so it has more time to like mush closed and all this kind of weird online mapping data stuff we're trying to munge together in our own platform.

[00:38:16] Ed Freyfogle: We had a very, very cool last, uh, cool talk last summer at Geo Mob in London. This guy, he works for, um, an agricultural technology company and check this out. So they fly the drone over a field and the way it works is so they're growing cabbages. And the supermarket will not take the cabbage if it's too big or if it's too small, because if it's too big, it'll clog up the, you know, the, the conveyor belt sort

[00:38:41] Mike: supermarket.

[00:38:41] It's not small destroying agriculture. Small.

[00:38:44] Ed Freyfogle: Right. So, so while, so, but if the, if the cabbage head is too small, then you know no one will buy it. Okay. So it needs to be exactly within sub parameters. So they fly the drone over the field as the cabbages are, are, you know, growing. And then they used [00:39:00] the, the photo analysis to target every individual head of cabbage and, and you, you know, do a prediction of like, this one is gonna, you know, hit the gross size or not.

[00:39:09] And, and then based on that they target, you know, how much water, how much fertilizer, how, you know, hype, hyper precise. I mean, it was amazing. Um, and, and it really was, you know, each individual, it was about taking all these different technologies and kind of chaining them together. Uh, and, and as a result, they able to be much, much more efficient in, in the watering and the use of, of fertilizers, pesticides, things like that.

[00:39:33] So it was a great, great talk. Um, and, and that's what I mean as an example of some of the things happening around geospatial. Um, I mean there are of course lots of consumer applications, but, but the real power is coming more and more in the combination with, um, you know, tools like Earth observation, you know, now it used to be in, you know, it used to be a very rare thing that you would get an aerial photograph of a place.

[00:39:54] Now you can basically target, you know, you can purchase a satellite photo of anywhere, you know, like put in your [00:40:00] credit card and have it immediately and start doing analysis on it and things like that. And so this is really quite cool. Yeah, this, there's a lot going, there's really

[00:40:07] Phil S: cool stuff happen in that space.

[00:40:08] I mean, specifically your cabbage thing. Um, I. It's really funny. I feel two halves of me kind of being pulled in two different directions With that, there's the kind of, I spend a lot of time on farms with farmers and they're just fed up with all the nonsense that happen to be put through with all the grants and all the kind of demand for tech and change, and supermarkets trying to low ball them at every single point.

[00:40:28] So they have to fire everyone and use massive machines and then everyone shouts at them for using too many emissions. Um, but then I also kind of feel the environmentalist that's like, let's be more efficient with everything that we do and get the best right. Um, you know, results. And also kinda like, yeah, the, the, the techie of just like, we could totally fix this with a drone and some algorithms.

[00:40:47] Like all of those things are competing in my head right now to be like, this is a genius idea and it's fucking stupid. Get it out of there. Fun in here

[00:40:54] Ed Freyfogle: right now. I mean, one of, one of the big use cases of geocoding is, as I said, is vehicle tracking. Yeah. [00:41:00] And a lot of that is cargo, cargo tracking. And you know, I mean, there's this amazing stat that like, you know, we, we produce, I don't, I don't know what it is.

[00:41:07] It's like two x the amount of food that the world needs. Right. It's just we throw half of it away. Yeah. Right. Because it, because it's harvested at the wrong time, or, you know, it rots on the way to the, into the store and all these kinds of things. So, Maybe by using all these different technologies along the way, we could be much more efficient and, um, yeah, for

[00:41:24] Phil S: sure.

[00:41:24] I mean the replacing, replacing whole field, spraying of pesticide with, um, drones, some of them by air, some of them are fondling along like some little early prototype styles, robot, but they're just kind of rolling around, spraying weeds that pop up cuz they can see 'em from space and drone instead of just spraying the whole damn thing and killing off the entire river and everything in it.

[00:41:45] Um, so yeah, that sort of stuff is definitely gonna help us, you know, continue to exist as a, on this planet,

[00:41:52] Ed Freyfogle: which is nice. I, I hope so. I mean, as I really, it is pretty fascinating to be a member of this industry to see, to see it [00:42:00] really come together and the, and the, the, the making the data available, making the devices, but have become much more cheap and much more robust.

[00:42:08] It's really cool. And, and, and then you add things like drones and you add aerial imagery and things like that to it. It's. It's, it's an exciting time in Geo special. Yeah. It feels

[00:42:18] Mike: like there will be no shortage of things to keep you busy for the, uh, immediate and far future, uh, ed, um, it's been super fantastic having you join us on the show today.

[00:42:27] Um, thank, thanks so much for being here. Um, before we let you go, I'm curious, where is the best place for people to find you online and where can they go to get started with open Cage?

[00:42:36] Ed Freyfogle: Well, anyone, anyone of course who needs geocoding should go to our website, which is, it's not open cage.com. It's open cage data.com.

[00:42:43] But if you just search for Open Cage geocoding, you'll, you'll find us. Um, yeah, if you're interested in geospatial stuff, come along to Geo Mob or, or listen to the podcast of geo mob.com. For me personally, my, my, my surname is, of course, is a bit difficult for spelling, but [00:43:00] let's go for it. My, my website is fry foal.com.

[00:43:02] Maybe you can get that in the show notes. Absolutely. Yeah. Um, in terms of social media, the best, um, best place with Maed, I've, I've fully made the switch to. Right on. Me too. Um, uh, also LinkedIn. You can, you can ping me on LinkedIn as well if it's, if it's, if you'd like. So exactly how I feel about LinkedIn.

[00:43:20] Anyone, anyone who wants to get in touch with.

[00:43:23] Phil S: Nice. So yeah, you'll, you'll notice when I'm starting to feel a bit broke because I'll start posting on LinkedIn more that the place I go to when I like, need a job or need to refresh my cv. I forget that people use that as a thing otherwise, but, um, yeah. Cool.

[00:43:35] Well I'll, I'll find you on Master On Cause I'm on there now doing a bit more over there. Um, it's good. You'll usually see me just post some really like, um, very short bullshit on Twitter, like some very little mini shit post and then mastered on, there's more space, so you actually get a thoughtful piece instead.

[00:43:49] I quite like that longer

[00:43:50] Ed Freyfogle: form. One, one quick note on that, you know, so Open Cage also has a mastered on account and we, we do a weekly thread about what we call geo [00:44:00] weirdness, so hashtag geo weirdness, where we talk about some of the. You know, we profile different countries and talk about all the different, you know, wackiness of, you know, enclaves and exclaves Absolutely.

[00:44:09] And border disputes. And, um, so actually this week for, for, for our British listeners in honor of the upcoming coronation, we had one about all the remaining British overseas territories like Pit, Karen Island and, and um, Gibraltar and uh, uh, places.

[00:44:27] Phil S: Nice. I'm following that right now. I am very interested in map

[00:44:30] Ed Freyfogle: weirdness.

[00:44:33] Yeah, no, we have quite a lot. We, we, we've done, I don't know, 20 different countries or so now every country has its own weird edge cases, so. Awesome. Thank you so much. Um, guys, it's been great to be on the show. Thanks for having us. Yeah, ed, thanks for joining.

[00:44:45] Mike: Cheers. Take care. Bye. Bye.