I had the good fortune of receiving an advance copy of Ken Auletta’s forthcoming book “Googled, The End of the World as We Know It“. It’s a fascinating read, one that raises a whole set of interesting dichotomies related to Google and their business practices. Contrast the fact that the Google business drives open and free access to data and intellectual property, so that the world becomes part of their corpus of data – yet they tightly guard their own IP in regards to how to navigate that data. Contrast that users and publishers who gave Google the insights to filter and search data are the ones who are then taxed to access that data set. Contrast Google’s move into layers beyond web sites (e.g., operating systems, web browsers) with their apparent belief that they won’t have issues stemming from walled gardens and tying. In Google we have a company that believes “Don’t be evil” is sufficient a promise for their users to trust their intentions, yet it is a company that have never articulated what they think is evil and what is not (Google.cn, anyone?).
There is a lot to think about in Auletta’s book – it’s a great read. When I began reading, I hoped for a prescriptive approach, a message about what Google should do, but instead Auletta provides the corporate history and identifies the challenging issues but leaves it to the reader to form a position on where they lead. In my case, the issue that it got me thinking most about was antitrust.
My bet is that in the coming few years Google is going to get hauled into an antitrust episode similar to what Microsoft went through a decade ago. Google’s business has grown to dominate navigation of the Internet. Matched with their incredibly powerful and distributed monetization engine, this power over navigation is going to run headlong into a regulator. I don’t know where (US or elsewhere) or when, but my bet is that it will happen sooner rather than later. And once it does happen, the antitrust process will again raise the thorny issue of whether regulation of some form is an effective tool in the fast-moving technology sector.
I was a witness against Microsoft in the remedy phase of its antitrust trial, and I still think a lot about whether to technology regulation works. I now believe the core position I advocated in the Microsoft trial was wrong. I don’t think government has a role in participating in technology design and I believe the past ten years have adequately illustrated that the pace of innovation and change will outrun any one company’s ability to monopolize a market. There’s no question in my mind that Microsoft still has a de facto monopoly on the market for operating systems. There’s also no question that the US and EU regulatory environment have constrained the company’s actions, mostly for the better. But the primary challenges for Microsoft have been from Google and, to a lesser extent, from Apple. Microsoft feels the heat today, but it is coming from Silicon Valley, not Brussels or Washington, and it would be feeling this heat no matter what had happened in the regulatory sphere. The EU’s decisions to unbundle parts of Windows did little good for RealNetworks or Netscape (which had been harmed by the bundling in the first place), and my guess is that Adobe’s Flash/ AIR and Mozilla’s Firefox would be thriving even if the EU had taken no action at all.
But if government isn’t effective at forward-looking technology regulation, what alternatives do we have? We can restrict regulation to instances where there is discernible harm (approach: compensate for past wrongs, don’t design for future ones) or stay out and let the market evolve (approach: accept the voracious appetite of these platforms because they’re temporary). But is there another path? What about a corporate statement of intent like Google’s “Don’t be evil”?
“Don’t be evil” resonated with me because it suggested that Google as a company would respect its users first and foremost and that its management would set boundaries on the naturally voracious appetite of its successful businesses.
In the famous cover letter in Google’s registration statement with the SEC before its IPO, its founders said: “Our goal is to develop services that significantly improve the lives of as many people as possible. In pursuing this goal, we may do things that we believe have a positive impact on the world, even if the near term financial returns are not obvious.” The statement suggests that there are a set of things that Google would not do. Yet as Auletta outlines, “don’t be evil” lacks forward looking intent, and most important it doesn’t outline what good might mean.
Nudge please …
Is there a third way — an alternative that places the company builders in a more active position? After almost two decades of development I believe many of the properties of the Internet have been documented and discussed, so why not distill these and use them as guideposts? I love reading and rereading works like the Stupid Network, or the Cluetrain Manifesto or the Cathedral and the Bazaar, or (something seasonal!) the Halloween Memo‘s. In these works, and others, there is mindset, an ethos or culture that is philosophically consistent with the medium. When I first heard “Don’t be evil” my assumption was that it, and by definition good, referred to that very ethos. What if we can unpack these principles, so that builders of the things that make up these internets can make explicit their intent and begin to establish a compact vs. a loose general statement of “goodness” that is subject to the constraint that “good” can be relative to the appetite of the platform? Regulation in a world of connected data, where the network effect of one platform helps form another, has much broader potential for unintended consequences. How we address these questions is going to affect the pace and direction of technology based innovation in our society. If forward looking regulation isn’t the answer, can companies themselves draw some lines in the sand, unpack what “don’t be evil” suggested, and nudge the market towards an architecture in which users, companies, and other participants in the open internet signal the terms or expectations they have.
Below is a draft list of principles. It is incomplete, I’m sure — I’m hoping others will help complete it — but after reading Auletta’s book and after thinking about this for a while I thought it would be worth laying out some thoughts in advance of another regulatory mess.
1. Think users
When you start to build something online the first thing you think about are users. You may well think about yourself — user #1 — and use your own workflow to intuit what others might find useful, but you start with users and I think you should end with the users. This is less of a principle and more of a rule of thumb, and a foundation for the other principles. It’s something I try to remind myself of constantly. In my experience with big and small companies this rule of thumb seems to hold constant. If the person who is running the shop you are working for doesn’t think about end users and / or doesn’t use your product, it’s time to move on. As Eric Raymond says you should treate your users as co-developers. Google is a highly user centric company for one of its scale, they stated this in the pre-ample to the IPO/s3 and they have managed to stay relatively user centric with few exceptions (Google.cn likely the most obvious, maybe the Book deal). Other companies — ie: Apple, Facebook — are less user centric. Working on the Internet is like social anthropology, you learn by participant observation — the practice of doing and building is how you learn. In making decisions about services like Google Voice, Beacon etc. users interest need to be where we start and where we end.
2. Respect the layers
In 2004 Richard Whitt, then at MCI, framed the argument for using the layer model to define communication policy. I find this very useful: it is consistent with the architecture of the internet, it articulates a clear separation of content from conduit, and it has the added benefit of been a useful visual representation of something that can be fairly abstract. Whitt’s key principle is that companies should respect the distinction between these layers. Whitt captures in a simple framework what is wrong with the cable companies or the cell carriers wanting to mediate or differentially price bits. It also helps to frame the potential problems that Side Wiki, or the iPhone or Google Voice, or Chrome presents (I’m struck by the irony that “respecting the layers” in the case of a browser translates into no features from the browser provider will be embedded into the chrome of the browser, calling the browser Chrome is suggestive of exactly what I dont want, ie Google specific Chrome!). All these products have the potential to violate the integrity of the layers, by blending the content and the applications layers. It would be convenient and simple to move on at this point, but its not that easy.
There are real user benefits to tight coupling (and the blurring of layers) in particular during the early stages of a product’s development. There were many standalone MP3 players on the market before the iPod. Yet it was the coupling of the iPod to iTunes and the set of business agreements that Apple embedded into iTunes that made that market take off (note that occurred eighteen months after the launch of the iPod). Same for the Kindle — coupling the device to Amazon’s store and to the wireless “Whispernet” service is what distinguishes it from countless other (mostly inferior) ebooks. But roll the movie forward: its now six and a half years after the launch of the coupled iTunes/iPod system. The device has evolved into a connected device that is coupled both to iTunes and AT&T and the store has evolved way beyond music. Somewhere in that evolution Apple started to trip over the layers. The lines between the layers became blurred and so did the lines between vendors, agents and users. Maybe it started with the DRM issue in iTunes, or maybe the network coupling which in turn resulted in the Google Voice issue. I’m not sure when it happened but it has happened and unless something changes its going to be more of problem, not less. Users, developers and companies need to demand clarity around the layers, and transparency into the business terms that bound the layers. As iTunes scales — to become what it is in essence a media browser — I believe the pressure to clarify these layers will increase. An example of where the layers have blurred without the feature creep /conflict is the search box in say the Firefox browser. Google is default, there is a transparent economic agreement that places them there and users can adjust and pick another default if they wish. One of the unique attributes of the internet is that the platform on which we build things is the very same as the one we use to “consume” those things (remember the thrill of “view source” in the browser). Given this recursive aspect of the medium, it is especially important to respect the layers. Things built on the Internet can them selves redefine the layers.
3. Transparency of business terms
When platform like Google, iTunes, Facebook, or Twitter gets to scale it rapidly forms a basis on which third parties can build businesses. Clarity around the business terms for inclusion in the platform and what drives promotion and monetization within the platform is vital to the long term sustainability of the underlying platform. It also reduces the cost of inclusion by standardizing the business interface into the platform. Adsense is a remarkable platform for monetization. The Google team did a masterful job of scaling a self service (read standardized) interface into their monetization system. The benefits of this have been written about at length yet aspects of the platform like “smart pricing” arent’t transparent. See this blog post from Google about smart pricing and some of the comments in the thread. They include: “My eCPM has tanked over the last few weeks and my earnings have dropped by more then half, yet my traffic is still steady. I’m lead to believe that I have been smart priced but with no information to tell me where or when”
Back in 2007 I ran a company called Fotolog. The majority of the monetization at Fotolog was via Google. One day our Google revenues fell by half. Our traffic hadn’t fallen and up to that point our Google revenue had been pretty stable. Something was definitely wrong, but we couldnt figure out what. We contacted our account rep at Google, who told us that there was a mistake on our revenue dashboard. After four days of revenues running at the same depressed level we were told we had been “smart priced”. Google would not offer us visibility in how this is measured and what is the competitive cluster against which you are being tested. That opacity made it very hard for Fotolog to know what to do. If you get smart priced you can end up having to re-organize your entire base of inventory all while groping to understand what is happening in the black box of Google. Google points out they don’t directly benefit from many of these changes in pricing (the advertisers do pay less per click), but Google does benefit from the increased liquidity in the market. As with Windows, there is little transparency in regards to the pricing within the platform and the economics. This in turn leaves a meaningful constituent on the sideline, unsatisfied or unclear about the terms of their business relationship with the platform. I would argue that smart pricing and a lack of transparency into how their monetization platform can be applied to social media is driving advertisers to services like Facebook’s new advertising platform.
Back to Apple. iTunes is as I outlined about a media browser — we think about it as an application because we can only access Apple stuff through it, a simple, yet profound design decision. Apple created this amazing experience that arguably worked because it was tightly coupled end to end, i.e, the experience stretched from the media through the software to the device. Then when the device became a phone, the coupling extended to the network (here in the US, AT&T). I remember two years ago I almost bricked my iPhone — Apple reset my iPhone to its birthstate — because I had enabled installing applications that weren’t “blessed” by Apple. My first thought was, “isn’t this my phone? what right does Apple have to control what I do with it, didn’t I buy it?” A couple of months ago, Apple blocked Google Voice’s iPhone application; two weeks ago Apple rejected someecards’ application into the app store while permitting access to a porn application (both were designated +17; one was satire, the other wasn’t). The issue here isn’t monopoly control, per se — Apple certainly does not have a monopoly on cell phones, nor AT&T on cell phone networks. The trouble is that there is little to no transparency into *why* these applications weren’t admitted into the app store. (someecards’ application did eventually make it over the bar; you can find it here.) Will Google Voice get accepted? Will Spotify?, Rdio? someecards? As with the Microsoft of yesteryear (which, among other ills, forbade disclosure of its relationships with PC makers), there is an opaqueness to the business principles that underlie the iTunes app store. This is a design decision that Apple has made and one that, so far anyway, users and developers have accepted. And, in my opinion, it is flawed. Ditto for Facebook. This past week, the terms for application developers were modified once again. A lot of creativity, effort, and money has been invested in Facebook applications — the platform needs a degree of stability and transparency for developers and users.
4. Data in, data out?
API’s are a corner stone to the emerging mesh of services that sit on top of and around platforms. The data flows from service providers should, where possible, be two way. Services that consume an API should publish one of their own. The data ownership issues among these services is going to become increasingly complex. I believe that users have the primary rights to their data and the applications that users select have a proxy right, as do other users who annotate and comment on the data set. If you accept that as a reasonable proposition, then it follows that service providers should have an obligation to let users export that data and also let other services providers “plug into” that data stream. The compact I outline above is meaningfully different to what some platforms offer today. Facebook asserts ownership rights over the data you place in its domain; in most cases the data is not exportable by the user or another service provider (e.g., I cannot export my Facebook pictures to Flickr, nor wire up my feed of pictures from Facebook to Twitter). Furthermore if I leave Facebook they still assert rights to my images. I know this is technically the easiest answer. Having to delete pictures that are now embedded in other people’s feed is a complex user experience but I think that’s what we should expect of these platforms. The problem is far simplier if you just link to things and then promote standards for interconnections. These standards exist today in the form of RSS, or Activity Streams — pick your flavor and let users move data from site to site and let users store and save their data.
5. Do what you do best, link to the rest
Jeff Jarvis’s moto for newsrooms applies to service providers as well. I believe the next stage of the web is going to be characterized by a set of loosely coupled services — services that share data — offering end users the ability to either opt for an end-to-end solution or the possibility of rolling their own in a specific domain where they have depth of interest, knowledge, data. The first step in this process is that real identity is becoming public and separable from the underlying platform (vs. private in, say The Facebook, or alias based in most earlier social networks). In the case of services like Facebook Connect and Twitter OAuth this not only simplifies the user experience but identity also pre-populates a social graph into the service in question. OAuth flows identity into a user’s web experience, vs. the disjointed efforts of the past. This is the starting point. We are now moving beyond identity into a whole set of services stitched together, by users. Companies of yesteryear, as they grew in scale, started to co-opt vertical services of the web into their domain (remember when AOL put a browser inside of its client, with the intention of “super-setting” the web). This was an extreme case — but it is not all that different from Facebook’s “integration” of email, providing a messaging system with no imap access, one sends me an email to my imap “email” account to tell me to check that I have a Facebook “email”. This approach wont scale for users. Kevin Marks, Marc Cantor, Jerry Michalski are some of the people who have been talking for years about an open stack. In the later half of this presentation Kevin outlines the emerging stack. I believe users will opt — over time — for best in class services vs. the walled garden roll it once approach.
Google search increasingly serves to narrow my experience on the web, rather than expand it. This is driven by a combination of pressure inherent in their business model to push page views within their domain vs. outside (think Yahoo Finance, Google Onebox etc.) and the evolution of an increasingly personalised search experience which in turn tends to feed back to me and amplify my existing biases — serving to narrow my perspective vs. broaden it. Auletta talked about this at the end of his book. He quotes Nick Carr: “They (Google) impose homogeneity on the Internet’s wild heterogeneity. As the tools and algorithms become more sophisticated and our online profiles more refined, the Internet will act increasingly as an incredibly sensitive feedback loop, constantly playing back to us, in amplified form, our existing preferences” Features like social search will only exacerbate this problem. This point is the more subtle side of the point above. I wrote a post a year or two ago about thinking of centres vs. wholes and networks vs. destinations. As the web of pages becomes a web of flow and streams the experience of the web is going widen again. You can see this in the data — the charts in distribution now post illustrate the shift that is taking place. As the visible — user facing — part of a web site becomes less important than the API’s and the myriad of ways that users access the underlying data, the web, and our experience of it, will widen, again.
I have outlined six broad principles that I believe can be applied as a design methodology for companies building services online today. They are inspired by others, a list of whom would be very long, I’m not going to attempt to document it, I will surely miss someone. Building companies on today’s internet is by definition an exercise in standing on the shoulders of giants. Internet standards from TCP/IP onward are the strong foundation of an architecture of participation. As users pick and choose which services they want to stitch together into their cloud, can companies build services based on these shared data sets in a manner that is consistent with the expectations we hold for the medium? The web has a grain to it and after 15 years of innovation we can begin to observe the outlines of that grain. We may not be able to always describe exactly what it is that makes something “web consistent” but we do know it when we see it.
The Microsoft antitrust trial is a case study in regulators acting as design architects. It didn’t work. Google’s “don’t be evil” mantra represents an alternative approach, one that is admirable in principle but lacking in specificity. I outline a third way here, one in which we as company creators coalesce around a set of principles saying what we aspire to do and not do, principles that will be visible in our words and our deeds. We can then nudge our own markets forward instead of the “helping hand” of government.