oreilly.comSafari Books Online.Conferences.


AddThis Social Bookmark Button

Inside Odeo with Evan Williams

by Richard Koman

When something gets as much big-time press as podcasting (to the point where Apple supports podcasts in the latest version of iTunes, released earlier this week; Apple, by the way, has no plans to charge for podcasts), either the hype machine is way out of control, or there really is something interesting going on. For Evan "Evhead" Williams, the creator of, podcasting is an exciting next step in the grassroots authoring movement he helped start. More than that, Williams is banking that podcasting is a really big next big thing.

He hooked up with Noah Glass--creator of the Audioblogger service that lets you create MP3s via phone--to start his new company, Odeo. The goal of Odeo is to make podcasting "easy enough for your mom." Odeo will allow users to find, listen, sync, and create podcasts via a good UI and super-easy tools.

That's a far cry from the current state of podcasting. Podcasting is somewhat easy to use as a consumer, once you've figured out which software you need, what feeds you want to subscribe to, how to subscribe, and how to copy them onto your iPod. Creation is a stickier enterprise. While most every PC these days comes with a microphone, it's not the highest quality way to go. Serious podcasters have to figure out their gear, their recording and editing software, licensing issues for music, difficulties in capturing phone conversations, and the technical details of creating the podcast-aware RSS feed.

And then there's the issue of content; just as with the Web and blogs, not all the content is particularly impressive. So there's plenty of room for Odeo to deliver some goodness and to stake a claim as the podcast service you'll tell your mom to use. I sat down for an interview with Williams at Medjool Cafe on San Francisco's Mission Street back in March. I wrote about Odeo for SiliconValleyWatcher at the time, but now that the site is almost ready for public beta (almost--not quite), we thought we'd bring you the full interview.

If Evan's description of Odeo catches your fancy, then you might want to visit the site and sign up as a (pre public) beta tester. Not all of the site capabilities described in this interview are enabled quite yet, such as the "create" tools, but what's there looks great. Here's more info about what they're up to...

Q: Why do you think podcasting is going to be huge?

Related Reading

iPod and iTunes Hacks
Tips and Tools for Ripping, Mixing and Burning
By Hadley Stern

A: I sort of came to the idea slowly, as I did with blogging; maybe not as slowly. Last fall, I had been commuting down to Google, and I had resisted getting an iPod for a long time because I didn't really like the idea of walking around listening to music, but since I had this long commute, I got an iPod with the idea of subscribing to and listening to books. At the same time, Noah, my cofounder in Odeo, had been running Audioblogger for a couple of years, and people had been recording audio on the web, which I found I never listened to, because there's just this little gray bar and ... it wasn't good on the Web, I felt.

But in talking to Noah and Biz Stone at Google we felt, well, if I'm syncing my iPod to get my books, why can't I listen to my blogs as well ... and that could be automated to create a TiVo-like experience. That was pretty compelling, and then Adam Curry and Dave Winer starting talking about podcasting and people started creating this content.

The reason I think this is significant is that there are so many hours when people are out in their cars. Citing the Long Tail has become a cliche already, but I can think of no better example than this. Think of all the hours people are driving. The odds that people want to listen to something in their cars other than what's on the radio at that exact moment--it seems obvious to me that that's the case.

The reason we focused on spoken word is because its really new. There's been a distribution channel for audio books, and is the clear leader in downloadable digital audio books. I love Audible--it's a great service. But I don't want to listen to just books any more than I want to read only books. I also like news and blogs and personal content. And that has never had a distribution medium in audio. If they're big enough, then maybe they can get on NPR, "This American Life" or something like that. Even before the Web, there were 'zines, but personal audio never has had a distribution medium. So there's a whole potential world of audio thats never had a distribution medium.

Q: Why doesn't audio work on the PC, but it does on an iPod?

A: It's interesting. I think there will be a substantial number of people listening to audio at the computer. People do listen to streaming radio. There's a difference when you get to spoken word, because that's less passive, but it's probably easier to listen to an hour-long show on the PC than a bunch of five-minute shows because there's not a mechanism to find them and play them all in a row. That gets back to the Audioblogger thing, coming across these things every once in a while and it's maybe 30 seconds and its maybe five minutes, that's a very disjointed experience and it doesn't work very well.

Podcasting is so nascent right now. Lots of people have done great work and its evolved much faster than, say, blogging. But it's not nearly something you can point your mom at--and that's what interests me: making it accessible and easy and enabling distribution to and from non-geeks.

odeo home

Q: Let's take a look at Odeo. Your page says "Listen, Sync, Create."

A: We're tackling three areas of functionality. The first is just listen, where we are aggregating all the audio feeds we can find. For every feed--we call it a channel--we create a page. We want to take as much audio on the Web as we can find and make it easy to find, and easy to point to, and easy to listen to. On every level, you can actually play the audio--there's a built-in player on the page, and there's metadata, descriptions, and comments. It's more than a directory, which some other people are building. We're actually aggregating the feeds. We're crawling them, we're displaying the content on the Web, without the user having to do anything else.

odeo show

Q: So each page is actually pulled out of the feed with software. No one's sitting there filing out the description.

A: No. People can edit their descriptions, but basically this is similar to a web-based aggregator. This is like Bloglines for audio. Every RSS item that comes in, we're creating a page for it, we're making it searchable, findable, etc. We're slicing and dicing the content in lots of different ways. And there's a social element, so you can see who else was subscribed to a channel, what else they're subscribed to, etc.

Q: Are you applying any selectivity to the feeds available? There's a certain breaking point to the old Yahoo directory idea of cataloging everything.

A: I think other people will create more of an edited version. We're taking more of a Google approach--find as much data as possible and then use community filtering and algorithms to help let whatever's interesting to you as a listener float to the top. And we have search built in now that's weighted to a variety of factors, but we'll want to build the collaborative filtering, ratings, whatever kind of mechanisms will be useful, but no limits on content.

Q: And you have tags.

A: Yes. Community-based tags. Anyone can add tags to a channel or a piece of audio if its not claimed. If it is, the owner can decide if other people can tag it. And we're extracting tags out of descriptions as well; if something exists as a tag, then we'll autotag it--most of the stuff coming in has tags.

odeo subscription

As you notice, every channel has a subscribe button and every piece of audio has an "Add to Queue" button. One of my fave things is you don't have to subscribe. That's one of the things I think will happen in general to RSS: more granularity. Subscribing at the feed level doesn't always make sense. And I think that makes even more sense with podcasts. You maybe want the Gilmor Gang with Adam Bosworth on it--maybe you don't want to listen to the hour-long show every week. If you want to subscribe its a one-click operation.

We have this client piece, Syncr, which we've built off of the iPodder source code and basically stripped a lot of stuff out, at least UI-wise. All your management is done on the Web. You never have to paste feeds into iPodder or use the directory component. You load it up, log in, and it checks the server and it syncs on a regular basis.

odeo sync r

If you want to download, you need to run this client. You don't have to run ours. This is doing the same thing as other readers, but we tried to make it really easy, but it's open--you could use anything. You never have to look at an RSS feed or copy and paste a URL. This grabs a personalized feed from Odeo, takes all these things, puts them on your computer, and creates a playlist in iTunes or Windows Media Player.

Q: So the create piece has been really, not that easy.

A: The thing that we think will work for most people is called Odeo Studio. It's designed to hit the sweet spot between being easy to use and being able to do interesting things. The most basic functionality--hit record, you're recording stuff. "Hello, this is Evan and Richard from Cafe Medjool in San Francisco." Hit playback, play it back. You can publish--it's in your channel.

If you want to do a little more you can add what we call Elements, which can be either text notes, or a more interesting element is audio, which you can upload straight from this interface or choose any you've already uploaded. Audio elements could be anything from your intro music, quotes from other shows, whatever it is. So then, when you hit record, you're recording and you play an element. When you play back, it will trigger the audio elements and wrap it up all together.

odeo studio

Q: How do you mix them in?

A: Basically you do it all live. When you're recording, anything you do is added to the track. Hit record then maybe you play your intro, you can trigger them at any time, you can load more while you're recording, set in points and out points. If you want to quote something, you can queue it up beforehand, you can loop it, etc., but you record it all live. Then when you're happy with the show, hit publish.

Q: Let's talk about revenue models.

A: OK. We're looking at three major things for revenue. The one that is probably the smallest but the soonest is on the hosting side. We expect the majority of content to be free, and we'll host for free up to a certain point, and that'll basically be based on how much demand it's driving. If you're Adam Curry and you're getting 50,000 downloads a day for a 40-minute show, we're probably not going to host that for free. We'll start charging for that; not sure how much. But that's sort of a proven thing. That's what's going to cost us a lot and its what people are already used to paying for. Hopefully we'll be able to serve media really efficiently.

Q: What about the case where people want to host it themselves rather than paying you?

A: That's fine, too. That's a good idea, actually. That's actually the original blogger model where we didn't even have hosting--it was just "send this to my server." That makes me a little hesitant, because for the small subset of people who want to do that, it can get pretty complicated, functionality-wise. But I would definitely look at that. I think its a good idea and we want to have as many options as possible. And a lot of people do have servers they can put stuff on. If we have a good creation tool, there shouldn't be any reason they need to use us to host.

Q: OK, hosting is one.

A: We don't see that as big business, we just see that as a way to cover some costs. The one I'm most excited about is premium content. That's where we were talking about Audible earlier. They've shown that people will pay for digital content that's not music. And I think that has a lot of potential in non-book form.

We don't expect there will be a lot of actual books on Odeo. But there's a lot of things I'd be willing to pay for in audio form, even if they're available for free in text form, and not pay $20 for them, but $1 or $5. If we're successful in creating a critical mass of listeners, then we can aggregate those payments and make it feasible to build a regular revenue stream for content creators without having to build payment mechanisms, without subscribers having to have an account at 100 different places.

Q: You see a revenue-sharing model with creators?

A: Yeah, I think of it more as a fee-based model than revenue sharing. We don't see ourselves as publishers, but more as a platform.

Q: How does the content creator get paid under this model?

A: They can decide what they want to charge, how they want to charge, by item, by subscription or whatever, and we'll have a fee structure. It will be a revenue share in that they only pay us when they make a sale and we take a fee for handling that interaction.

Q: A pay-per-customer model, as opposed to "pay us to use our platform and handle the sales yourself?"

A: Right, the former. We want as much content as possible. and we expect the majority of our content to be free. And we want good content because that will drive up customers. But if the content does well then we'll take a reasonable fee.

Q: A thing with podcasts is that they're all over the map from NPR to--just like blogs, right? Complete inanity, badly recorded.

A: Definitely.

Q: So presumably people would pay for quality, both quality content and quality production values.

A: That's the hope. And that's completely speculative. I have no doubt some people would pay; how big a business that is, I have no idea. It's like a lot of content businesses; there's some content some people are willing to pay for, there's more that's free and people are willing to put out there for free, and there's advertising as well. That's somewhat proven. Live365 has built a business partially around advertising in audio. We should have good numbers on subscribers and there's a lot of products for which audio advertising should work pretty well.

Q: So this is like advertising in the stream like a radio ad?

A: Right. And it could be done in a few different ways. We're not going to auto-insert ads without users' permission. We want to create a marketplace for that, where advertisers could come and say, "OK, I want to advertise on this show and here's my ad" and they can upload their ad and then the content creator can say, "OK cool, I'll accept that and I'll import it when I'm doing my show," or maybe have a tool to auto-append it or something like that. We're not going to have a sales force selling those ads, or a production team producing ads. It's more like creating an opportunity for those folks to get together.

Q: What about an AdSense-like matching of ads and content?

A: I think that'd be cool. It sounds really hard. We may be able to do some simple things based on tags and descriptions. I'm not sure how well that would work if it's totally automated. It might be a filtering mechanism, and then advertisers can choose channels.

Q: So I as a user don't have to worry about you populating my content with your ads?

A: No. No. That doesn't appeal to me at all. We want the content creators to be happy. There may be an option to do that, maybe that's instead of paying for hosting or maybe that's as a shared revenue, but that's--you'd get something for that.

Richard Koman is a freelancer writer and editor based in Sonoma County, California. He works on SiliconValleyWatcher, ZDNet blogs, and is a regular contributor to the O'Reilly Network.

Return to the Mac DevCenter