The Attention Operating System
Microsoft’s Yahoo takeover, whether successful short or long-term, marks an historic change in Microsoft’s perception of its role at the center of the computing universe. Certainly Google’s rise has focused the Redmond mind on the task it must confront, not just with Google’s advertising dominance but also its dagger to the heart of Microsoft’s crown jewels, Office. At its most basic level, the Yahoo deal allows Microsoft to clone Google Apps and blunt the hemorrhaging of a new Net-aware generation away from the current hardware bound Office.
We can argue whether the forced merger of MSN/Live and Yahoo’s services will be easy or difficult, but those who predict distraction and brain drain should remember that Ray Ozzie was the central factor in the IBM/Lotus acquisition that allowed continuity and pragmatics to preside over what turned out to be a successful combination. Notes and Domino achieved a critical mass in resources and seats that it needed to blunt the Y2K Exchange challenge, and IBM bought the time it needed to invest in open source and build out a platform for Global Services to ride on.
Similarly, Microsoft can use the Yahoo seats to deliver an 80-20 version of Office while financing the transition with improved response for brand advertising hung off of Yahoo media properties and rich media services such as Flickr. Banner ads don’t attack Google’s advertiser/relevance stranglehold, but they do pay for the migration to an on-demand Office clone and give Microsoft time to confront the real Google threat: control of an Internet operating system.
Where Windows achieved its lock-in based on control of the hardware base, an Internet operating system requires a different equation: the engagement of the user, or put another way, getting and retaining the attention of the user. This social contract is one where Microsoft has a fighting chance of success, and significantly, one where Google has its challenges. Recent attempts by Google to use data collected in Gmail as social graph gestures with Google Reader shared feeds underline how difficult it is to use data collected under one social contract to seed an emerging one.
Imagine how we would have felt if we were told that the data Google collects in Gmail that informs the suggested links in our private email discussions would be made available to other companies for their marketing purposes. Or the government for census data or the insurance companies for purposes of calculating liabilities for pre-existing conditions. But turn that around and ask the question differently: would users be willing to release their attention data in return for discounts, access to information, membership in special interest groups, and so on? An inversion where the harvesting of behavioral data is released by the user as what I’ve dubbed gestures.
Attention was first proposed in 2004 by Technorati founder Dave Sifry and me as an XML specification called attention.xml. The notion was that the digital breadcrumbs we emit around the network could be captured and transmitted as a simple signature of behavior: who, what, and for how long. In RSS, this breaks down into the feed, the individual post or item, and the length of time spent on the page. In other words, the attention of the user. A clickstream recorder such as the one we released in October 2005, or in fact, the recordings left by us as we browse services from Google, Yahoo, and every other site, are aggregated and processed based on the implicit understanding of the value of the service. What permission do you give us in return for the “free” services that we provide?
Gestures invert the permission model, moving past the inference algorithms of anonymous users to the direct messages of intent. Social networks represent the first cut at this paradigm: I’ll tell the service not only what my behavior is but who my friends are. The service then aggregates the gestures of my friends and streams it to me. So far so good. I can be invited to events, send gestures such as RSVPs that then are broadcast back to my affinity group, and so on. Or via Twitter, I can follow, and generate timely updates on what is happening, what is important to me, and what I want to know more about. But the behavior, the attention data, is not yet being used to filter the gesture stream. The attention data of my friends, when aggregated anonymously, enormously improves the signal to noise of my information system.
Combining attention with the social graph of validated gestures produces a new and extremely valuable economy. An economy where the price of admission is the credibility of the social contract. To reiterate, Gmail already knows what I’m interested in, what I am doing, and especially what my friends and business associates are doing via email, IM, blogging, and Twittering, all of which I am piping into Gmail and Gtalk where it is archived and increasingly made available via Google Reader. Getting permission to use that data is what is hard, not the infrastructure.
So Google has the necessary data but not the required permission. The rise of social constructs such as Facebook and Twitter provide an intermediary for such social contracts. By agreeing upfront to an mapping of behavioral signals (attention) to the explicit social relationships of gestures, services can be constructed that respect the user’s wishes and mine the anonymous aggregate data of the affinity groups that represent targeted qualified streams of intent. Vendors who want access to the efficiency of this marketplace have to respect the signals sent by those users: I am interested in this, don’t spam me with that, support these ideas that I support, introduce yourself to me via those people I respect, and so on.
Microsoft’s opportunity is that Google has changed the playing field from the one Microsoft used to dominate. It’s not so much that Microsoft is no longer the black hat but that in a world of choice, the white hat belongs to the one or ones who respect the user. That is, assuming there is a credible choice for these so-called free services. In fact, the battle is for ownership of the attention operating system, and the only dominant position possible is one of equal choice between two or more credible services.
It’s an odd kind of mutually assured destruction, where Microsoft and Google each are constrained from “winning” by the knowledge that if we can’t be reasonably confident of an alternative, we won’t invest our clicks in any one place. Microsoft can afford to invest in the Yahoo cloud, and can’t afford to destroy that company’s trust relationship with its customers. In other words, they can’t afford not to do this deal. And they can’t afford not to build an Internet OS that equally respects Windows, Linux, and OS/X, as Silverlight does. It won’t be easy for either company to keep going down this road, but luckily they have no choice as long as we do.
February 12th, 2008 at 3:45 am
How does the Google Reader issue differ to Microsoft doing the exact same thing with Live Favorites/Live Spaces? If you click Share in Live Favorites, it goes to your Live Space by default.
http://img.skitch.com/20080212-ut7rp1c4rju29ktkaprp6j1k3.jpg
Are Microsoft breaking the contract too?
Plus it isn’t just limited to Spaces. By default, Windows Live Messenger contacts can see items in your “Whats New” feed (aka News Feed) and vice versa. Sound familiar huh?
Activity from your Space shows up in other users feeds by default. You have to explicitly turn this stuff off, just like in Google Reader. You also have to specifically state that no Messenger contacts can get access to this attention data.
Its not just Favorites though. SkyDrive, the hosted storage service from Microsoft allows you to keep files in the cloud with two settings - private or public. By default, those marked as public show up in the feeds of all Spaces/Messenger contacts.
How does any of this differ to the Google situation? This is a clear effort to reverse engineer the Messenger social graph for the benefit of the Spaces social graph, when in reality for many people there is a difference between the two.
February 12th, 2008 at 6:10 am
Steve I have a problem with your obsession with Google and the reader stuff - I still don’t get it.
I got why facebook was bad with Beacon immediately and so did most of the people I told - they were monitoring what you did on other sites and telling your friends about it. This to me is clearly about permission.
I think I must have a misunderstanding about the gmail / reader issue because to me it is. Gmail has the ability to autofill email addresses for you (I love that ability) and it does this by storing details of anyone you send an email to in the contacts database. Google Reader has a shared function that allows you to publish a public shared feed of interesting blog posts. Until recently the shared feed was public but you had to send the unique URL to people, now anyone that you have sent an email to who also uses gmail can access your feed.
So to me the problem is that Google had a feature that was public but because it was complex to use people assumed privacy - Google made it easier to use and those people that had assumed it was private now found that because the assumption was invalid they had now published all of their shared items to all the people they had emailed.
I do not think this is a privacy issues with Google sharing your data with someone else - I think it is a lack of understanding by Google that people had made this assumption and a lack of understanding about how some people were using the shared item feature.
I’ve explained this story to others and can’t get anyone who cares about it in the same way they did about Beacon and Facebook.
I am afraid that perhaps I have missed something in the story and I look forward to finding out the whole story.
However if I haven’t can you please concede that this is should be considered a lack of judgement by the Google Reader team on how people are using the feature than it is a general lack of respect for permission by Google.
I love the Gillmor gang and your posts but your obsession with the Google Reader issues really is beginning to ruin some great discussions.
Again I will accept my mistake and you can ignore the above if Google is now showing Pepsi and Levis all the details of my shared item activity.
February 12th, 2008 at 11:31 am
Is there a Silverlight runtime for Linux? You cannot say that Silverlight equally respects Windows, Linux, and OS/X until the *tools* to build Silverlight apps are available on all three operating systems. Android apps will make the leap from mobile devices to browser plug-ins and RIA frameworks before Silverlight gains any traction. Just because Microsoft is talking to you and Google is not does not change the reality that Google’s platform is much more attractive to developers.
February 12th, 2008 at 4:41 pm
@ Scott, Microsoft themselves are not creating a Silverlight runtime for Linux, though I believe the architecture is in place for others to do so. I am not an expert on these matters though.
@ AI - “I love the Gillmor gang and your posts but your obsession with the Google Reader issues really is beginning to ruin some great discussions.”
I agree completely. I am far from a Google apologist but Steve seems to be using this sole (and not that significant) issue as a means to completely devalue any merit the company has when it comes to the user contract, and as a base for his new-found Microsoft love. Some of the comments have been eyebrow raising to say the least. I mean, *Microsoft* of all companies! It must be the Ray Ozzie appreciation clouding judgement. I kid, I kid.
Also, Shared items are shown to contacts you have actually chatted with in Google Talk, and not solely Gmail “contacts”. While I am not excusing the lack of foresight on the Google team, there is a subtle difference between the two, especially in that Gmail has the auto-add algorithm.
February 13th, 2008 at 4:24 pm
Wills,
You seem to argue that Microsoft Live sharing features equate to Google violating the user contract. Apples and oranges. Are you saying that Microsoft did violate the contract? Then you have an analogy. Try again.
If you don’t care about user contracts changing under you, then I don’t think you’ll get much of an argument. A silent gesture, yes . . .
Scott / Wills,
Novell is partnering with Microsoft to build a Silverlight for Linux. The plan for non-Windows support was envisioned from the beginning. The distinction of who writes the code is immaterial.
February 14th, 2008 at 4:12 am
Hey Steve - any answers?
February 18th, 2008 at 3:08 pm
I just set up Google Reader. It asked a million questions before any sharing happened and it was very, very clear to me what my options were. So I am left wondering what all the brouhaha is about. Sure they shouldn’t have automatically shared anything without asking, but it seems that has been fixed. The questions in my mind are: How long did they take to fix it? How much pressure had to be applied before they fixed it?
Google made a boo-boo in this instance but it pales in comparison to Beacon. And speaking as someone who grew up in the Microsoft era, I find it incredible to think that anyone believes that Google has dug themselves a deeper hole than Microsoft when it comes to having the trust of their users. I’m sure that 98% of Google users still think of them as the company with the motto “do no evil”, who gives them lots of cool free software. Contrast that to people’s sentiments about Micro$oft.
Maybe MS learned from Hailstorm and are now privacy’s staunchest advocates. But maybe Google learned from their stumble too.
February 19th, 2008 at 4:01 am
I agree Google Reader only seems to share items that you have marked as shared anyway.
I wish Steve would give us some input on this - I guess there is something we are missing - otherwise I don’t understand why he keeps harping on about this!
February 20th, 2008 at 5:45 am
Lets not forget Microsoft handing over user data to the Chinese government, hand over fist. Google was the only search company to put up a fight.
February 24th, 2008 at 5:29 pm
[…] the arrival of the Attention Economy. Gillmore recently explained the concept of attention in a post analyzing the market positions of major players: “Attention was first proposed in 2004 by […]
February 29th, 2008 at 12:22 pm
I’m just waking up to the outrageous terms of service for Google’s cloud apps as described here: http://blogs.zdnet.com/Greenbaum/?p=151
Once again, lack of respect for the user. Maybe Microsoft can steal a march on Google after all.
March 25th, 2008 at 9:12 am
there is no free lunch … what is a kilowatt of electricity to the typical user? 10 CDs played? 3 hours of web browsing? 1/100 of a shopping cart transaction?
the observation concerning the importance of time, however, is very good …
time not location has become the resource that needs to be measured more consistently with the cost of access and/or bandwidth use … there is a there there, there and there … and if it isn’t someone will provide it … there
additional variables :: a balance between what is *believed* to be “privacy” (delimited by the potential or cost of ID theft — perhaps in a SLA or EULA or an insurance provision relating to same - for valuation or liability limitation purposes) and what can be characterized as either “piracy” or “marketing” of intangibles (whether by statute such as the Patent Act or the Copyright Act or even consent decree such as the ASCAP & BMI arrangement for performance of music or some other as yet to be determined attribution for value creation) … this can apply to aesthetics or functionality of the network, the application or transport layer, packet flow and the underlying protocols to support efficient AND/OR effective “trusted transactions”
pricing access in terms of the cost of bandwidth or the computational expense to maintain access-based rules (to drive a particular but related transaction[s]) is bounded by how to gain the “trust” of the user (could be as “insignificant” as trusting a 5 dollar bill is indeed authentic, based on observation - right or wrong, or as detailed as verification from a certification authority for both parties to a given exchange) and more significantly how to get the user to pay (payment may also be subsidized by advertising or subscription or some other commitment of value in exchange for value to be returned to the provider) …
the enabler that is lost in all of the discussion are the means and practices of how quickly payment options have been deployed and how quickly transactions can be had in real time or close enough to real time … will technology move as fast as payment (another potential variable) and can there be observations of present value if the windows for expected economic activity be reliably predicted while the rest of the content to be sold later has been paid for (catalog that was once newly released material)
payment providers can compete with network providers in an observable sense >>> what is the complexity attributed to consummating a given transaction? what about, in terms of computation and by extension bandwidth or time …
that being said, it is quite instructive that any return on investment for a given intangible requires better measurements or transparency, not only for the advertisers or aggregators, but for the end-user as well … time deficits exist for all market participants, real or perceived
recognition and trust are fungible and time-driven models simply enable even higher valuations for given network use … especially in a time-deficit economy
GREAT PIECE!!