(this is page 114 of 121)





Latest

Lots of people have been going ga-ga over the last few days about "Zooomr," a Flickr copy that supposedly adds a couple of more features and leaves Flickr "in need of catching up" according to Mike Arrington.

I don't even want to go into the comparison between an industrial strength app capable of handling 3 million users and 100 million photos, and a hack done by a single person (if you have any doubt look for a great presentation on how Flickr scales at Cal's site, or just try Zooomer now that TechCrunch and others have sent 30-50K visitors over there). What I will focus on is David Berlind's piece: "'Trapped data' raises switching barrier to newer, sometimes better services." In it he argues that as we use online services that require our data, we have to watch the hidden lock-in when a new, "better" service comes around the corner.

Now, there are a few things that you can criticize Flickr for. But the roach motel effect is not one of them. In fact, I would argue that of all of the data-centric sites out there, Flickr is the absolute best at having a great, great, open data policy. When we started building Tabblo, we spent 6 weeks using Flickr as our backend and outside of some sluggish API calls, it was just amazing to see how much richness Flickr provided in its API, and how well thought-out it was.

But it doesn't stop there. Months later, when we sat down to integrate Flickr into Tabblo, we were again surprised at how rich the facilities are for integrating metadata from Flickr (tags, contacts, privacy settings). We were also floored by how easy the Flickr guys make it, and quite frankly, because of this, anyone launching a photo-related site today that doesn't provide Flickr integration off the bat can't be such a prodigy.

Berlind is right to point out that Webshots has a piss-poor policy for exporting data (as in, you can't) but this is also true of Shutterfly, OFoto, and Snapfish. The fact is that too many of these Generation-1 photo sites still believe that holding the high-res images hostage will drive more business through the print engines. Given how painful it is to upload high-res assets through today's asymmetrical broadband pipes, this is a really bad thing to do to your users.

Recently at a conference, I asked an employee from one of these large photo sites why they didn't just clone some part of the Flickr API as a goodwill gesture and use it to position against their other Gen-1 competitors. She told me that "Emily from Des Moines doesn't care about APIs, or high-res images." And that may in fact be true of Emily, but it certainly isn't the case with all of those new services (like us) that think they can fill some needs not currently being addressed.

I'm all for picking on Webshots, and Ofoto on their lack of openness. But Flickr? Those guys should be given a medal where open data (and metadata) policies are concerned.

Comments Permalink

On Etech

Just came back from Etech where I failed to have a single "wow" moment. The closest I think I came was during the demo of the multi-touch interface the first day of the show. Overall, I left San Diego feeling like something was lacking.

Don't get me wrong, there is plenty to like about Etech as a venue. I got to catch up with Doc (more on him later) which is always great fun. I had lunch with Dave Sifry, who I'd only known through the blogosphere, and found him to be quite an inspirational guy (and like Doc a very decent closet shutterbug). I attended a scaling webapps fast-and-cheap tutorial put together by Cal Henderson from flickr which was great geek porn and perhaps the most honest, well thought-out presentation on the architecture of a large website that I have ever attended. And, of course, Joel is always entertaining enough to be worth paying attention to even if he's only half-right most of the time.

As I thought about why I found the contents of the conference somewhat lacking, it occurred to me that the blogosphere may be in fact responsible for the lack of "wow" moments. After all, most of the presenters at the conference are prolific bloggers, so most of the big ideas get out well ahead of their talks and company announcements. Back in the late 1990s at a conference like PC Forum, you were hearing these things for the first time (perhaps this is what is driving the interest behind the un-conferences which seem to place all of the emphasis on the attendees).

But upon further reflection it struck me that there was something else missing. The big theme of the conference was supposed to be "The Attention Economy" and the presentations, company pitches, etc. were cast in that mold. And yet, despite the fact that I am an avid listener of the Gillmor Gang, I now have no idea what the attention economy is all about. For a while, I thought I did: I thought the attention economy was all about the stream of metadata that came from using web applications. From clickstreams to search terms, we spend virtually all of our time online generating this metadata, and it often seems frustrating that it gets trapped at whatever property we happen to create it on.

But it turns out that the attention economy is about more than this metadata. It's also about money, and time, and intelligence augmentation, and gestures. It's about Web 2.0 APIs, and mash-ups. It's about event calendars that you use to schedule events you wish could take place. It's about distributed identity management. And most of all, it seems to be a suspension of disbelief around the need for business models that deviate from the AdWords template.

On the plane back from San Diego I was thinking about how this was what really bothered me about the conference, the fact that attention seemed like a funky euphemism for eyeballs instead of a synonym for metadata when to my great delight I ran into this piece that Doc posted on the need for an Intention Economy instead of an Attention one. He just nails it, most of all because he puts the user at the center of the value proposition. I think that this is what was missing from most of the Etech company presentations-- good old product management that focuses on what the user gets out of all of this (I saw in this post a funny term for it "Etech gets Ninged").

An example: Adam has been trying to work on a better type of aggregator. He is starting from a clear customer pain and trying to define a product around it. It's really as much about attention as anything else from the conference but because he's started from the customer pain, it feels a lot more concrete and tangible.

Comments Permalink

All Mushed Up

Adam, who is out at Mashup Camp, has been trying to get me to think about mashups for a while now. I think more than anything it's been the fact that he was willing to get into a tin can and fly 3,000 miles to an ad-hoc "camp" that has me finally getting over my "yeah but we did this 10 years ago with Perl scripts" bigotry and really trying to suss this out.

Before I get to where mashups click for me though, I should mention that it's nice to see that even while inhaling that west coast vibe, Adam has stopped to think about the fact that with the exception of Amazon, source component providers like Google, Ebay, and Craig's List don't have a real clear economic incentive to continue providing their data and compute cycles for free beyond a certain yet-to-be-determined point. This is a very valid and important observation which Peter Rip picks up in a nice post on the challenges associated with building a mashup-based business.

Now one might argue that allowing for core functionality to be exported serves a real long tail purpose. People who mix up Google Maps with anything else are perhaps the best example of this-- after all, while one could argue that the Google Maps team may have seen apps like Housing Maps coming, there is certainly no way they saw the Jacktracker (an intensely enjoyable site for 24 fans). I think the long tail argument is good (as is Eric von Hippel's related user-led innovation argument) but I think that mashups provide these big source service providers an even bigger benefit: the ability to attract an audience for their core functionality that is way out of the swath inhabited by their average user: the geeks.

Geeks are amongst the more discerning and most marketing-proof people you can find out on the web today. They're also among the most passionate, and with the emergence of the blogosphere, they are definitely the most vocal. Businesses have been built on the hypothesis that understanding geeks really matters. And yet, trying to build an application that has high geek appeal and that also serves a large market of "regular" users can be a bad idea for one of two reasons: either the application already exists and has won their hearts and minds (think photos and Flickr, bookmarks and del.icio.us) or it is just too far out there to have any sort of "regular" person appeal.

Exposing core functionality in the form of an API that facilitates mashups cuts this Gordian Knot in a productive way. It makes a potential service provider focus on what's really important and differentiable with an influential and discerning crowd. And at the same time, it lets the product team focus a more tightly integrated web application on the core constituency.

And so, here's what's been most useful to me about trying to get my head around the impact of mashups: thinking through the hypothetical of when geeks might want to mash up parts of what we're working on. It encourages clarity (in terms of what it is that is unique) and enforces modularity (so that we can easily export this core functionality without going crazy). And most importantly, it gives us hope of finding a nice way to serve two equally important (but very different in terms of needs) audiences.

Comments Permalink

As we struggle to get Tabblo into beta, Ned sent me a very relevant piece from Brad Horowitz-- "Creators, Synthesizers, and Consumers"-- on social sites, participation, and the industry's move to embracing the unwitting blogger.

Horowitz starts the post with this chart to show that in fact very people people have to be "active" in order for a community site to be of significant value. His argument is that this 1-10x-100x ratio holds across different sites and he uses both Yahoo Groups (a great aggregation of many communities of all types and sizes) and Wikipedia as examples of the ratio holding. He argues that this winnowing we see as someone moves from lurker to creator actually helps to reduce noise in the community which I don't agree with (who hasn't been part of a group where the most vocal can end up being the most disruptive?) but then also argues that he is all for removing barriers that preclude more active participation.

The first example he gives is one where the production experience is so lightweight (posting individual pictures to flickr) that it's a very easy and seamless shift from consuming to producing. This is interesting but I'd argue that for many worthwhile activities that live naturally in a social software site, the production experience shouldn't (or couldn't) be sufficiently simplified to make the transition so easy (researching a wikipedia topic, remixing podcasts, contributing code to an open source project, etc.) and what becomes really interesting to me are attempts to break the 1-10-100 ratio in those cases, especially around the 1-10 lurker-> synthesizer transition where I think there is so much locked up communal value.

Brad then makes the case for properly instrumenting a social site such that it encourages "implict creation" and creates "public artifacts" as a byproducts of what the users are doing in the system. This is similar to Dan Bricklin's old Cornucopia of the Commons argument or Tim O'Reilly's Architecture of Participation and is dead on as a starting point for trying to harness the potential creative investment of all of those community lurkers.

I prefer using "unwitting blogger" because there is a bumbling and serendipitous tone to unwitting that I think fits better with how most people who engage with a site suddenly discover that they have undergone a state change up the pyramid. It's almost never a conscious decision on the part of the user but after it's done, you'll find it very hard to put that particular genie back in the bottle.

Unfortunately, this dynamic seems like a really hard thing to get right. It takes a really rare mix of tool-building, community management, and perhaps most importantly (and most obviously missing from most of today's social sites), a really deep understanding of how people learn and what kind of rewards are necessary to get them to invest one more incremental unit of effort in working their way up the pyramid. As we move from easy production experiences to richer ones where more creativity is encouraged, finding the right mix is the challenge that faces us all.

[ As an aside, it is so nice to see a bigwig from the largest social software experimenter out there blogging about this stuff openly. It speaks volumes to how good it is to be working in the consumer web today, especially when you compare it to say, another "consumer" industry: consumer electronics where I had an opportunity to play five years ago. No matter how hard I try, I just can't imagine one of the big cheeses at Philips posting on whether "this iPod thing is really gonna go anywhere." ]

Comments Permalink

Marc Hedlund over at O'Reilly Radar had an interesting piece about how smart kids, when left to their own devices in a new environment, will independently come up with a set of good development practices for Web 2.0. Adam Green followed up with a post about how some of these practices aren't so new after all since he was using them almost a decade ago and Andover.net (both pre and post the Slashdot acquisition).

While I think Adam is right that the best of the Web 1 companies were implementing some of these practices back in the 1990s, it is much more the norm now than it used to be. Part of this comes from the maturing of the problem domain-- we are all that much more educated now than we were back then about how to think through building web applications. But the biggest contributor to this brave new way of doing things in my mind is transparency of commodity infrastructure.

Let me use an example: back in 1998 if you were building a web-based startup, you were probably running on Solaris/SPARC and using an Oracle database. You were also likely to be running on some sort of a Java servlet engine (though there were exceptions, this was again the leading edge). This huge apparatus usually required at least 1 of the following: DBA, sys-admin, release manager, and build manager-- nevermind all of the consultants and vendor people that it took to solve problems that arose from trying to get everything working together.

Fast forward to 2005. Anyone still using Solaris/SPARC for web apps is either a moron or a depressed Sun shareholder. MySQL and Postgres are now considered "enterprise-grade," and if you should be so masochistic as to still want to do Java development on the app-tier, you've got Tomcat, Jetty, and even JBOSS available to you on your platform of choice.

Now here is the key: every piece of infrastructure is free, has a thriving online community that can help you with issues better than the vendors ever did, and perhaps most importantly, can scale down to run on almost any type of laptop.

This last piece is what I found was missing from Marc's post: the fact that in the Brave New World, every developer can get to have the entire stack on his own machine. The value of this should not be underestimated. It allows for features to be prototyped and tested much more quickly. If you are running across different platforms on the dev machines (Linux, OSX, and Windows) it enforces good discipline in the build system and in the modularity of the code. The platform differences will also surface bugs which would not immediately show up in a more homogenous environment (for example, concurrency is a great thing to suss out by deploying your app across platforms).

When we started building Tabblo, I insisted that we be able to build the full stack across different flavors of Linux (our deploy platform), OS X, and Windows. Given that we have a bunch of C code in the application, this proved particularly painful at the outset in the Makefiles and Linux/OSX differences. Moving it to Windows meant dealing with a bit of a Rube Goldberg that leveraged VMware for piece of the app and the portability of tools like MySQL and Python for another piece. Having taken that pain up front however, I do not think that there is anyone at work today who regrets the investment. We can all work completely independently even while completely offline despite the fact that our deploy infrastructure requires a dozen machines running together in a data center.

This was just not possible last decade. We owe it to the transparency of the open source infrastructure that we can do this now across a whole set of components that used to be the domain of experts and alchemists. Anyone who doesn't take advantage of it now is doomed to slow release cycles, the need for a full QA team on hand, and a staff-imposed level of overhead that is tough to build a business out from under.

What a great Brave New World.

Comments Permalink