(this is page 106 of 121)





Latest

Let 1,000 flowers bloom

I like output based business models. It's no secret that there is something special and extra nice about being at that point where bits get turned into atoms and then charging for the experience. It's simple, it's clean, and it leverages all sort of good things about computers (customization), the Internet (a channel), and technology (mass customization). Whether it's custom clothing, books-on-demand, or some of the other interesting things we've got planned at Tabblo, the era of mass customization is definitely here to stay.

Which is why it's been a little depressing to make the rounds in Silicon Valley over the last 12 months– at the epicenter of the Google AdWords inspired haze– and see people look blankly at me when I get going about the pleasure of actually making things. The world there has turned into a hammer made of eyeballs and just about everything looks like ad inventory. Today it looks like Yahoo is proving that even in virtual goods, the laws of supply and demand take effect– and when there is not enough demand (and too much supply), the market corrects. Is it the beginning of the end for this new mini Web 2.0 bubble? Doubtful, mostly because advertising is not going anywhere but online (though it may be quite slimmed its life there). But it is nice to see that we may get some attention paid to two other types of business models in the coming months: those that are output based and those that are subscription based. There are a lot of talented folks at the big media companies that have spent far too much good brainpower "enhancing ad inventory" or "maximizing click-through rates" and it is going to be nice to see these guys get into another groove for a while.

Case in point: Moo.com and their mass-customized businesscards, launched just today. A really nice, really light-weight application for repurposing your Flickr stream for a simple (and compelling) product. Photo-like businesscards aren't for everyone, but the experience of leveraging the investment made into Flickr for something completely physical will only get better over time and it's good to see Moo taking this first step.

Also, kudos to Flickr/Yahoo for letting this type of creative experimentation bloom on their platform. Because it has been non-core to their ad business, they've let folks like Moo (and us) experiment on top of their APIs without the onerous requirement that a revenue-sharing infrastructure might impose at this early stage. In my mind, this will be the greatest legacy of the online advertising boom– it has seeded the fields for what comes after it in much the same way that the PC software boom (and its associated hardware advances/commoditization) laid the groundwork for the open source movement and for cheap scale-out datacenter infrastructure.

And when the time is right, I've no doubt that we'll be able to give back to the folks that let us into the playground in an economically meaningful way.

Comments Permalink

Marshall over at TechCrunch has an interesting post on people online getting paid to participate on social media sites. This is a timely debate now for a couple of reasons:

First, the whole Netscape/Digg thing (with Jason Calcanis from Netscape offering to pay the top social bookmarkers) kicked off a lively debate on whether this could be a viable means to build momentum online that surfaced a very intriguing possibility. Perhaps it is because social news is such a narrow application, but I think that Jason is right in assuming that a few people properly incented can suck the oxygen out of competitors' sites and create a lot of momentum quickly. The more interesting question in my mind is whether this same tactic can be used for deeper/richer applications– that is, could you build a YouTube by paying the top contributors? An eBay? I tend to be skeptical of this, but if it is indeed possible, it may give all of these hungry VCs with their trigger-happy capital fingers an opportunity to put a lot of money to work fast.

Second, the debate is especially timely because of how many great producers of content there are out there in the world these days. Chris Anderson beats this to death in his Long Tail, so I will just contain myself to one observation. Whenever I fly across the continent, I bring along my video iPod chock full of content. I've got bought episodes of The Office, Lost, etc. but I've also go a number of video podcasts. Last Saturday while flying back from SF, I was struck by how much I enjoyed two particular pieces of content: The Office, and Ze Frank's The Show (maybe it's all about the simple noun names). As I crammed in 10 3 minute episodes, I realized for the first time I was looking at content that I would pay a premium subscription for. Now, if I had started on Ze Frank by being asked to pay, I very much doubt I would have– but now that I am hooked, I should be able to pay for him to keep working.

Micro-payments and micro-content and the holy grail that people have been chasing probably since long before Web 1.0– but for some reason, it feels like we may be getting there now. Sure Anderson is right in pointing to the tools and the aggregators in his book, but I wonder if there isn't something else at work– some sort of major shift in how producers/makers actually do stuff. There have been echoes of this in recent bestsellers like Friedman's flat world book or Richard Florida's creative class work, but I don't think anyone has actually come out and spelled out the fact that we're in the middle of a huge transition in terms of defining what "work" is, how it gets done, and how one gets compensated for it. Even Om is getting in on the fun with a great new blog concept on what the "web worker" is about. Taken to it's limit case, one could imagine a much more immersive experience (a la Stephen Levy's new piece on WoW) where work gets done in much the same way that WoW is played, by loosely coupled but tightly-knit guilds that take things happen one project at a time, sometimes for immediate material gain and sometimes for fun.

We definitely do live in interesting times.

Comments Permalink

Good computer science canon teaches you that the specifics of the computer language you are using are at best "implementation details" and that if you grasp the key core components of computation (indirection, recursion, modularity, etc.) and are a good "computer scientist," it's just a matter of time before you're up to speed in anything from COBOL to OCaml. If this is true though, it's curious to see how many people have turned language choice into a more heated issue that the Red Sox versus Yankees one and how often it keeps coming up in flames and debates.

At the end of the day, I too would like to believe the theory that we are all just huge brains floating in jars full of jelly wired with electrodes and that external stimuli like syntax, core language constructs, whether there is an explicit compile step, static typing, etc. can just be adapted to by moving a few of the electrodes around. Unfortunately, experience just doesn't bear it out.

Which is why I found Luke Plant's tongue-in-cheek piece about how learning languages like Python and Haskell are making him less of a productive programmer such an entertaining read. For some reason, reading his C# code example made me think of all of those heinous years of programming in Java when I would be left with really ugly constructs like anonymous-inner classes-as-adapters or a horrific introspection API for faking late-binding code (and the fact that early Java heads would talk about how "elegant" these solutions were did not help my confidence at the time either). It's been two years since I've had to do that, and I can honestly say that I will probably never use Java again in any kind of serious project, or C# or any of the other similarly inspired languages (unless forced to under threat of starvation).

The reason is simple: working full-time in Python (and Javascript to some degree) rewires those brain electrodes in a really positive way (if your goal is fast development and deferring design decisions as long as possible which it should be for any web application developer; perhaps not so true for avionics engineers) and it would be both painful and counterproductive to go back at this point. Others have written extensively about this but the basic argument is that with an highly expressive, consistent, low-overhead language, you get a ton more time to think through the real problem. Conversely, another way to put it that is more in tune with my particular circumstances these days is that you need a lot less time to experiment and prototype.

So I feel for you Luke– I think if I had to still be writing the cruft that they've got you coding at work, I would also be less productive, but probably mostly from chewing my fingers down to stubs out of frustration. And to all of you computer science bigots with the "I can program in abacus if I have to" mentality– you know, the guys who say the One True Language is LISP even though the only time they've typed more than two parentheses in a row was when they were dabbling in ASCII art or who talk really fast about Turing-complete languages assuming that there is a 1-to-1 exchange rate between insight and speed of talking– to those guys, I look forward to getting you jobs with some of our competitors.

Comments Permalink

In the context of recent blogposts about pageviews being a silly metric, Google trying to entice its users into doing work on their behalf, and social bookmarkers getting paid to do what they used to do for free, Doc Searls has an important post on the coming age of the individual-in-the-driver's-seat when it comes to various different spheres of computing. Specifically, he makes the case for all of the web service providers giving up on Roach Motel 2.0 (where Roach Motel 1.0 was about keeping your data hostage, 2.0 is about trapping your metadata) and even cites the Tabblo-Flickr ecosystem as an example of the way in which data/metadata sharing can benefit both the user and the companies building stuff.

I am definitely not smart enough to follow all of the intricacies of the Attention Trust model or the Gesture Bank (around how the user can keep control of his metadata) but I did want to add one observation to Doc's argument: the reason why metadata is so important to the "vendors" he describes is because they are all wed to an advertising model that implies monetizing eyeballs. Though this "media model" is often shrouded in all sorts of false complexities, the core levers behind it are very simple: you need increasing returns in the ability to get more eyeballs (which is why raising switching costs is so important) and a deeper understanding of those eyeballs than any of the competition who might also have access to them. With those two things you get more inventory and have the ability to charge more for it.

While this model continues to work, I find it pretty difficult to believe that anyone is going to make significant inroads putting the user in charge of anything that so core to it's continued success. In our case, our primary business is around two forms of value capture: converting bits to atoms (specialty print) and offering the user premium services (subscriptions). In both of these cases the value proposition gets made directly to the enduser and so it behooves us to make sure we make them as happy as we've made Doc with his creative investment of 11K photos on Flickr and his tabblos on Tabblo.

At the end of the day, companies, like people, are motivated by incentives and constrained by the rules of the game they choose to play and I don't think that even Doc would fault them for that. It's only when the advertising model starts to evolve to take account of this (and I mean more than just AdWords-like evolution though I am by no means a media expert) that things will really start to change for the user on a mass scale.

One other related note on lock-in: people (investors, partners, etc.) have been asking since day one of Tabblo: where is your lock-in? I used to have a lame answer about network effects and user-generated content that I would give, rattling off comparable efforts like eBay's and MySpace's. I don't do that anymore. And the reason is because I came to the conclusion that looking for this type of lock-in is similar to adopting a revenue model that has a high probability of putting you at odds with your users. Instead, we prefer to focus on execution, and specifically on locking product development to user demands. Better integration with Flickr? Full-res downloads? We're happy to oblige under the working hypothesis that the talented creative class on the Net responds well to companies that take care of their needs and that– to Doc's point– cycles spent there create a deeper relationship between company and users– which may be the only kind of lock-in we'll all be left with at the end of the day.

Comments Permalink

Ned has jumped into the fray on web application frameworks and this reminded me that just about a year ago when we were getting Tabblo off the ground, my head was spinning (or more specifically my laptop was) with every web framework known to man to see which one we'd begin with. At the time my thought was that we would use one of the frameworks to get started and then abandon it as we needed to scale the site to something beyond a prototype/early alpha.

So it's been a pleasant surprise to see that we've been able to stick with Django for the last year and even through a couple of massive spikes in traffic (something I am fairly sure that we would not have been able to do with some of the other contenders I considered so kudos to the Django guys).

However, Ned's post brought back all the memories of reading reams of commentary on all sides of web frameworks and passionate arguments around how "game changing" this or that was, and how software development was never going to be the same again. He does a good job of debunking most of this as the crap, so instead I just wanted to mention one of the greatest benefits I think we've seen from using a "fringy" framework like Django.

If you get passionate about your tools– passionate enough to try new things, get into the fray on the discussion, and always be looking critically at everything that comes down the pike, you're probably the type of engineer who thinks hard about tradeoffs and the "whys" behind decisions which means you are most likely also to do very well coming up with the right answers to some of the harder (softer) questions that Ned has on his post. In other words, getting people who are passionate about their web frameworks, especially when these frameworks are not part of the mainstream "career advancing" canon, is a great way to pre-select for the types of people who will thrive and do really well in small team environments. This isn't always true– and interestingly enough, as the tool/framework gains acceptance (something which should be considered a "good thing"), it's value as a preselector rapidly decreases (what Java monkey doesn't "totally love Ruby on Rails" these days?).

It reminds me of a story that I've heard third-hand about the way that Steve Jobs did recruiting at NEXT. When asked what he was looking for in hardware engineers, he supposedly said "engineers that understand surface mounted board assembly," and we he was asked why that specifically, his response was "it's easy: the best hardware engineers all want to work with surface mounted components, so that is why we want them." When I first heard this story I thought, what an idiot! but working in the Django ecosystem has taught me that this type of self selection can be much more beneficial to a startup than URL dispatchers, ORMs or any of the other stuff that comes with a web framework.

Comments Permalink