I attended MobileCamp San Francisco (mobile site) yesterday. What a great event, it was a BarCamp – a user driven un-conference with no spectators, where everyone is a participant. Although BarCamps are user generated events it takes the efforts of dedicated volunteers to pull off an event like this. Thanks Indira, Ritwik, Andy and Alexi, I know you worked your butts off to make MobileCampSF the success it was. Thanks also to Nokia, whose sponsorship provided great food and a perfect venue in the beautiful and architecturally significant Swedish American Hall. It was great to finally meet Mike and David and so many of my readers. I attended and participated in as many sessions as I could and here are my impressions.
Photos by Ritwik Dey. More MobileCampSF photos
First was PlayMegaphone.com‘s, Dan Albritton. Megaphone has a really cool series of products built around the unique concept of controlling digital signage with mobile phones using voice and/or keypad. They have a deal with MTV to use the huge MTV video screen in New York City’s Times Square to present multi-player games that passersby are encouraged to participate in, by using their phones. We played two games at MobileCamp. The first was Shoot’em where you steer your phone around the big screen zapping other players. The game was a lot of fun and after Dan gave the winner of the first round her choice of a Nokia N95 or N800, everyone wanted to play. The other game Noisy, Noisy Hippos was even better. You’re are a hippopotamus and grab objects flying across the screen above you by yelling “Grab!” into your phone. It must have been hilarious to watch a roomful of techies yelling “grab, grab, grab…” although I was too busy trying to win an N95 to really notice. About 10 devices were given away to winners. I didn’t win anything but came close, oh well. MegaPhone seems to be doing really well with their idea. They have done games for various sponsors all around the world on big screens. The concept scales down to smaller screens in stores and bars. They did games at the Palm Cento promotion and Dan showed a clip of Palm CEO Ed Colligan enthusiastically playing. Dan sees the product as an alternative to SMS campaigns which he categorized as “Do something boring and we will SPAM you.” Instead Megaphone offers fun with game winners getting their prizes by SMS or a voice message. The server runs on a laptop and supports 30 users and scales to thousands of users with multiple servers using Amazon’s EC2 rent by the hour servers.
Next I participated in a small group led by Nelz, brainstorming mobile security and privacy issues. Combining social networking with mobile technology that follows us everywhere we go and has the ability to report our location with an accuracy of 50 meters raises huge privacy issues. How can we use the technology without subjecting ourselves to surveillance by government, employers, credit bureaus and individual stalkers? We generally agreed that mobile users do not currently have a good way of knowing what information will be shared and with whom. We came up with ideas including: campaigns to educate users of the danger of revealing personal information including mobile phone numbers to third parties and encouraging users to demand that information not be shared without explicit opt-in. We also came up ideas for new business models based on enhancing user security like anonymising mobile proxies, disposable phone and SMS numbers, and granting a “Seal of Good Privacy” to vendors that meet certain criteria for protecting user identifiable identification.
Bebo‘s Jordy Mont-Reynaud did a presentation titled “What’s The Killer App in Mobile Social Networking? ” Jordy defined 4 elements of a killer mobile social application
- Reach – Make it ubiquitous. Support lots of carriers. exclusivity doesn’t scale. People won’t participate if they can only interact with friends who have the same carrier.
- User Experience – The ability to pushing messages to users is a key success factor, mobile users are busy doing other things, they don’t want to be constantly checking for messages.
- Technolgy – SMS and WAP are preferred over downloadable applications for scalability, rapid development cycles and easy deployment.
- Control Cost to user – use bandwidth sparingly. Seek a “high fun to byte ratio.” Most users don’t have data bundles, and only some will have text bundles. Users expect social networking to be free based on desktop experience but are willing to pay if it makes sense. Be up front with users about costs and encourage them to have data and messaging bundles.
Bebo is a 1.5 years old web based social network and member of OpenSocial, that claims to be the 3rd largest social network internationally. Last month Bebo launched a mobile version on Orange UK which includes a special 3 pound/month unlimited text and web bundle for Bebo users. Bebo will be launching off-portal on all networks by year end.
I caught the tail end of Alexis Rondeau’s (Semapedia) presentation on QR codes. Alexis mentioned the importance of open standards and pointed out that QR and Datamatrix, the two open formats for 2d codes, have another advantage beyond being open and royalty free compared with proprietary codes like Shotcode. QR and Datamatrix encode a URL which the readers pass directly to the browser. The proprietary codes take the user to the vendor’s site for a lookup or billing before redirecting to the final destination, which has a performance impact and privacy implications. As he ended the presentation, Alex bemoaned the lack of phones with bundled 2D code readers especially in the US. He mentioned that Nokia was now bundling a reader with several phones. Then he asked participant’s who hadn’t won a phone or tablet to raise their hands and handed each of them a new N95. So I got my phone, although I felt I kind of cheated as I only participated in part of the session. I’ll be posting more about the phone after I’ve explored it some more starting with the QR code reader.
Mike Rowehl lead an exploration of Maemo, the OS running on the Nokia N770 and N800 tablets. We all got to play with a couple of prototypes of the next model, the N810, which is smaller than it’s predecessors but keeps the same great 800 x 480 4.1″ screen. It features a slide out QWERTY keyboard and GPS. I could definitely see using one of these to blog on the road. Nokia has promised to release 5 revisions of the Nxxx tablets and of Maemo so the N810 is the third in the series. Interestingly the browser on the N810 is a “lite” version of Mozilla that is NOT Minimo. Maemo is completely Open Source OS including the kernel, X windows server and window manager. The C++ and Java development environment runs on a Linux desktop but Mike codes mostly in Python right on the device. There are already 400-500 Maemo applications. The big hope in the Maemo communitiy is that Nokia will release a Maemo phone. Already Maemo has been ported to FIC’s Neo, the OpenMoko phone. Alexis tried to give away N800’s at this session but everyone present had already won either an N95 or N800 so he moved on to one of the other sessions to visit Nokia’s largess on some folks.
The final session I attended was my favorite and certainly the liveliest. Dave Harper and Nigel Choi did a very polished and fascinating presentation on how Vodafone and Novarra are causing serious grief for mobile developers, off-port vendors of downloadable content and users. The issue, which I’ve blogged about before, is that Voda has made all web access pass through Novarra’s transcoder by default. The transcoder replaces the mobile browser’s User Agent header with a generic desktop User Agent making it impossible to tell one phone from another. Sites that do browser detection to deliver targeted mobile or desktop content are fooled into sending the desktop version to the transcoder. Novarra does a programatic reformatting to create a bowdlerized mobile site. Although Novarra claims that the result is a better user experience, Nigel showed a series of slides demonstrating how the transcoder completely breaks the Wall Street Journal’s site. Instead of the WSJ’s well designed mobile site with full access to articles, Novarra delivers a site that forces users to scroll through 5 pages of irrelevant sidebars and headers before they reach the story. But it gets worse, the desktop WSJ site requires a subscription and users, after scrolling down 5 screens, are told to login “at the top of the page” which of course is 5 screens back. You can see Nigel’s presentation here.
Novarra, in addition to claiming that their transcoded sites are better than purpose built mobile sites, has come up with some numbers to the effect that since enabling the transcoder Vodafone’s content sales on it’s portal are up. This makes sense. If the transcoder breaks off-portal web sites and content downloads, users have no choice but to use the Voda portal. Very sneaky. Vodafone has extended transcoding to Ireland, Spain and Italy using various other services in addition to Novarra. Voodafone UK offers a whitelist where content providers can request that their site not be transcoded. There is no common whitelist for all Vodafone national operators, content providers to apply separately to each Vodafone division. Developers are fighting back but there’s a sense of powerlessness against the might of a giant corporation. Giving up is not an option. Some of the actions that have been started and need support:
- Luca Passani’s petition is the central rallying point. If you haven’t signed it yet, please do.
- Pressure is being but on the W3C and government regulators. This is a slow process. Vodafone and Novarra are both W3C members and are seeking to get the W3c to justify their anti-competitive and anti-standards behavior.
Some brainstorming lead to the idea of seeing if the transcoder can be detected from it’s headers. Sites could then but up a message saying the content had been degraded and urging users to complain to Vodafone. Interesting idea, I wish I could sniff those headers but that’s not possible from this side of the pond.
MobileCamp is coming to New York City this Saturday, November 10. It looks like it’s full, but there’s a waitlist. If you live near New York, give it a shot.
Pingback: MobileCamp San Francisco
Cool, great to hear more perspective on this. I should also point out dotMobi’s note on this topic:
http://dev.mobi/files/ContentTransformation_consultative.html
I personally am following this debate, even if I’m not actively involved in it myself. I myself am surprised about the User-Agent business — I’d think that sites want to know that it’s talking to a proxy instead of wanting it to pretend it’s a phone. I can understand some special cases (ads) that also want to know what the original user agent was. In that sense I myself don’t see what’s so wrong with Novarra’s behavior. Heh, Google’s own transcoder doesn’t tell you anything about the original device.
I personally hope the combination of existing mechanisms are sufficient to signal that “I am handling adaptation myself thank you — please just let me talk directly to the phone.” The OpenWave issue you cite sounds, well, like an OpenWave issue — I suppose it doesn’t like being unable to transform the WML into HTML.
But that is a problem in practice, yes, and I hear some sentiment here that transcoders are going to have to do whatever it takes to make this work as expected in practice. For example, transcoders may need to make a request with the original UA, and if it looks good, go ahead and pass it through, and resort to a second request if needed.
I can’t speak for Sean but I do not think he quite meant it in the sense it was understood. I mostly seem him concerned about cases where the site sends back markup that might work quite poorly on the target device (e.g. trip up on bugs in the browser, sending way too much data). I hear some broad agreement that ultimately the origin server should have control.
Thanks for your comment Sean, and thanks for the link to the W3C draft. I hadn’t read it before.
I should have made it clear that the objection is not to transcoding. It’s to Vodafone unilaterally transcoding all web content without user opt-in and with no advance warning to the development community. In particular, they are reformatting mobile sites and even http messages sent by non-browser mobile applications.
It was generally agreed that transcoding serves a useful purpose in making content usable on handsets that would otherwise not be able to render it. Google was specifically mentioned several times as a transcoder that a user could choose if he or she wanted to view a desktop site on their phone. Google was also praised for doing the right thing in using the the meta tag rel=”alternate” media=”handheld” to detect mobile sites and pass them through un-modified.
I agree that if Requirements 5 and 6 of the draft as proposed are adopted and Vodafone and Novarra follow them, it would resolve the issue in a way that I as a mobile web developer could live with going forward. Of course it doesn’t do anything for legacy sites and applications which are broken by Novarra/Vodafone unilaterally changing the User Agent.
“Requirement 5. Origin servers and proxies must be able to identify the actual identity of components of the delivery context, including (other) proxies and browsers.”
There is currently no standard for this. Novarra is using x-Device-User-Agent which AFAIK is not part of any standard. This requirement needs to specify a new required standard header for sending the original user agent.
“Requirement 6. Origin servers must be able to prohibit any kind of transformation of its content.”
Again the draft doesn’t specify how origin servers are supposed to do this, which I guess reflects that it’s a draft and hence unfinished. The transcoder user by Vodafone Ireland doesn’t currently respect “Cache-Control: No-Transform” according to comments on the Betavine site although the UK one does. I’ve also noticed a side effect of using “Cache-Control: No-Transform”. Openwave gateways give an error and will not display the page if wml is sent with this header. For this reason, I’d rather see a new header that identifies a site as mobile AND prohibits it’s transformation rather than using the existing “Cache-Control: No-Transform”
Working with the W3C on this issue is certainly necessary. However the Vodafone transcoders are breaking sites and services now. Standards bodies like the W3C move slowly. The effected content providers need a solution now, and rallying developers and users to put pressure on Vodafone seems to be the one way to possibly achieve that.
The text you quoted (“Vodafone and Novarra are both W3C members and are seeking to get the W3c to justify their anti-competitive and anti-standards behavior.”) is mine and it is not a direct quote of anything David or Nigel said. It is based on Sean Patterson of Novarra’s widely reported message to the W3C’s Best Practices Working Group where he claims “A well-designed content transformation server can do a better job of following the mobile best practices than a
human author, especially when taking into account the capabilities of the many different mobile devices. The result will be a more consistent, uniform experience.”
Perhaps Sean really believes that but personally I think it’s pure BS designed to confuse the issue and justify Novarra’s practice of rendering transcoded desktop versions of sites that offer purpose-built mobile web sites in spite of the wishes of both content providers and users.
Dennis
Pingback: Mike Rowehl: This is Mobility » Blog Archive » MobileCampSF
I noted the comment here: “Vodafone and Novarra are both W3C members and are seeking to get the W3c to justify their anti-competitive and anti-standards behavior.”
I wonder if people had actually read the current W3C draft on this topic?
http://www.w3.org/TR/2007/WD-ct-landscape-20071025/
It asserts, among other things, that the origin server must be able to control the transcoding proxy’s behavior. Not sure what people were led to believe, but within the context of the W3C, Novarra and Vodafone are pushing for exactly the kind of control that developers seemed to be clamoring for here.
I am glad to see folks were thinking about standards-based mechanisms by which elements of delivery chain can communicate and work together as desired. indeed, standards help here, and I do believe the W3C remains a good forum for hashing these out. Indeed — standards are already “on the books” here. The Novarra (and Google) proxies identify themselves in the User-Agent header, yes!
To get to the point: there are even existing mechanisms in HTTP for telling proxies to not transform: Cache-Control: no-transform
I hope that came up as well. It sounds like there was much angst rhetoric about ‘evil’ companies here (from a familiar source), instead of focus on actual working solutions. Focusing on Novarra is misguided since they are by no means the only transcoding proxy (Google is quite ‘guilty’ of the same behavior). I’m sorry to hear someone was harping on this. It’s up to informed developers like those at BarCamp to understand, debate and promote existing standards to solve this problem for real sites, now, rather than spend time writing petitions and so forth.