July 28, 2009

microsoft's settlement proposal

(Note: This is my personal opinion and does not necessarily reflect Mozilla's position or any formal statement from Mozilla. I expect that Mozilla will make some kind of public comment on this topic, but this is definitely not that.)

On Friday Microsoft released a statement and a number of documents that they hope will lead to a settlement with the European Commission with regards to the ongoing legal case against Microsoft for violating EU anti-competition statutes.

The first collection of documents, on "Browser Choice", directly addresses the EC case and, if accepted by the EC, would presumably become the legally binding settlement with the EC.

The second document group called "Interoperability" is scoped beyond just the EC and the EU and is not intended, as far as I can tell, to be a part of any binding settlement with the EC.

When Mozilla became a participant in all of this, Mitchell Baker started a series of blog posts to develop a set of principles that Mozilla believed should be addressed by any remedies in the case. The principle discussions were not about specific remedies themselves, rather they were about the goals that we though any remedies should meet.

As I read through the Microsoft settlement proposal, it looks to me like Microsoft has at least made a nod to each of Mozilla's principles and I find that encouraging.

But I suspect there's two ways to read this. The casual reading, as if it was a press release or a blog post, may suggest that Microsoft has responded favorably to most of the concerns that most people voicing concerns (including Mozilla) laid out. From that reading, things probably look pretty good.

There's potentially another way to read this, though, as an excruciatingly precise legal document. Reading it in that light might elicit a different reaction -- it did for me and I'm not even a lawyer (though I presume to be able to think like one.)

Assume for a minute that Microsoft's exact language here was codified as the legal settlement document with the EC. What might lawyers on either side be able to assert or defend based on the exact language in the document. How do things look in that light?

I've tried to convey some of both readings in my reactions and I encourage those of you with the interest to try reading it both ways as well and to share your thoughts here in comments.

I structured this blog post by trying to connect each of the principles that Mitchell described with the part of Microsoft's settlement proposal that most closely aligns with that principle and I'm interested 1) whether or not you think there's really alignment between the two, and 2) whether the remedies Microsoft is proposing would, in your view, effectively satisfy each Mozilla principle.

From Mitchell's EC Principles: Synthesis blog post:

Of the various principles I proposed, the ones that get the strongest positive response are those that protect the choices people have already made or are trying to make. These are outlined in principles 1: Respecting Previous Choice and 2: Windows Must Not Provide a Technical Advantage to IE.

Let's take the first one first. Windows cannot subvert a person's choice of an alternative browser. Here's how Mitchell described the principle:

Once a person has chosen Firefox or Opera or another browser this choice should be respected. Neither Windows nor IE should use the presence of IE to encourage or promote a return to IE, or to automatically open a different web browser than that which the user has selected. Otherwise, the monopoly presence of Windows on 90+% of the world's personal computers means that people are forced to choose alternative products over and over again.

Mitchell went on to describe some specific area of concern:

  • Use of IE for operating system purposes cannot bleed into web browsing
  • IE must close after OS purposes complete
  • IE may not ask to become the default browser or make itself the default browser except in specified legitimate circumstances, like perhaps when a person downloads IE separately from Windows or from a Windows update

It will be useful to identify the ways in which Microsoft products -- including Office, as suggesting in a previous comment -- lead people to IE, or open IE as the browser even when another browser has been selected as the default. Feel free to add them here or let me know through other channels.

And here's the relevant text from Microsoft's settlement proposal:

1. (1) Microsoft will make available a mechanism in Windows Client PC Operating Systems within the European Economic Area (EEA) that enables OEMs and end users to turn Internet Explorer off and on.
1. (3) Within Microsoft's PC Productivity Applications [Microsoft Office 2007] distributed in the EEA, Microsoft shall not include any icons, links or short-cuts or provide any other means to start a download or installation of a Microsoft web browser. Microsoft shall not use Windows Update to offer any new version of a Microsoft web browser to users within the EEA unless Internet Explorer is turned on on the user's computer.... Windows Client PC Operating Systems shall enable end users to choose their preferred default web browser. Windows Client PC Operating Systems, including Windows Update, and Internet Explorer, shall not override the user's choice of default web browser.

So how does that sound to you? Will that put an end to Microsoft's ability to use Windows to undo or otherwise subvert user's choice of browser?

Here's what I see in this part of the Microsoft proposal:

Microsoft will provide users a switch to turn IE off at which point it disappears from the system and won't be turned back on by Office 2007 or Windows Update. That sounds good, but I don't think it will have much practical impact. For this to be meaningful, I think Microsoft needs to to do one of two things. One, they could make the IE off switch a public API so that it can be offered as a checkbox in browser installers ("Would you like Opera to replace Internet Explorer?") where it might actually get used some. Or, two, they could change the language to say they will not use Windows Update to ask users to make IE the default browser if the user has already made another browser the default. Microsoft has a long history of using Windows Update to undo user choice and they're basically saying here that they'll continue those practices unless IE is fully turned off. So, in my view, it needs to be really easy to turn off, or they need to extend the language to cover the case where IE's on but not the default browser.

Microsoft agrees to not advertise IE, either through links, icons, or shortcuts, from within Office 2007. That's a good start but I think it doesn't go far enough. What about Microsoft programs other than Office 2007 (I read that Office 2010 is just around the corner, for example,) and what about non-Microsoft programs that call IE directly when they really should be calling the default browser? If Microsoft wants this provision of their settlement proposal to matter, I think they need to cover more than just Office 2007.

The second principle that Mitchell articulated was Windows can't provide a technical advantage to IE.

Windows has provided technical advantages to IE through techniques such as those listed below. This list is not meant to be conclusive. It is meant to illustrate the range of ways in which Windows can and has made it difficult for other browsers to provide a competitive experience on Windows.
  • Making information available to IE before or differently than that information is available to others.
  • Making it difficult for other browsers to access browsing information stored in Windows, thus making migration and syncing painful for users and difficult for other browser makers to implement well. This includes information such as formats and metadata related to IE favorites, website passwords, and website cookies.
  • IE use of undocumented Windows APIs.
  • Providing APIs to IE available to Windows developers as part of the "Windows" API. As a result applications developed by third party developers can send URLs directly to IE rather than to the browser the user has selected as his or her choice.
  • Requiring the use of IE to use the Windows update service. (Microsoft appears to have phased out this practice, or to have provided alternatives. I include it as an illustration of the ways Microsoft has, and could again, use Windows to damage competition in the browser space.)

The ubiquity of Windows brings this artificial competitive advantage for IE to almost every single person using a personal operating system. Redressing this setting will help refocus competition on the merits of the browser itself.

And here's what Microsoft is offering the EC:

1. (1) Microsoft will ensure that if Internet Explorer is turned off, then (i) it can only be turned on through user action specifically aimed at turning on Internet Explorer; (ii) the user interface cannot be called upon by applications; and (iii) no icons, links or shortcuts or any other means will appear within Windows to start a download or installation of Internet Explorer
2. (14) Microsoft will ensure that all the Windows APIs on which Internet Explorer relies are disclosed in a complete, accurate and timely manner, so that non-Microsoft web browser suppliers would not be at a competitive disadvantage compared to Microsoft when designing a web browser for Windows. For purposes of this paragraph 14, Internet Explorer consists of the code that is distributed separately from Windows and trademarked or marketed as Internet Explorer.
3. (17) For Windows Vista and successors, Microsoft shall ensure that access and the full functioning of the Windows Update online service (currently available at www.update.microsoft.com) is not dependent on the use of the Internet Explorer user interface.

What do you think? Does that go far enough to mitigating the IE and Windows integration technical advantages?

I personally think this, on the face of it, represents reasonable steps in the right direction, but I'm bothered again by the fact that IE simply won't be turned off all that often so applications will continue to launch IE at which point IE will attempt to undo the user's default browser choice.

The API commitment sounds strong to me, so long as one assumes that the APIs are scoped in such a way that they were actually usable by browsers from other vendors. I'm also a bit concerned about their definition of "Timely Manner" which is "as soon as Microsoft has developed a sufficiently stable 'beta' testing version of Windows (including Service Packs and Updates) and has made this implementation available to third parties for testing purposes for the first time." There are two issues I'm concerned about here. First, is the first beta really early enough for third parties to take advantage or to alert Microsoft about concerns with the API that could negatively impact browsers from other vendors. The second issue is that I do not see a provision for the situation where there's an API that's new to a new Windows release but isn't yet utilized by Internet Explorer. If the requirement is to disclose only APIs that IE is using at the time of the Windows beta, what happens if the API isn't used at the time of the Windows release but IE later takes advantage of that API?

Windows Update is already pretty well separated from Internet Explorer with the stand-alone Automatic Updates program, but if I'm reading this right, Microsoft is saying that the online version will also not depend on IE. I'm a little unclear though, because they're not saying it won't require the IE rendering engine, just that it won't require the IE user interface. A more explicit "Windows Update online service at www.update.microsoft.com will work in Firefox, Safari, Chrome, and Opera" would have made me feel a lot better here.

The third Mozilla principle was Windows must enable people to choose other browsers

Mitchell put it like this:

How much distribution advantage can Windows provide to IE? If one answers "as much as Microsoft wants with no limits" then this principle wouldn't be implemented. If one answers "Windows can't provide any distribution advantage to IE" then one would likely end up supporting a remedy that requires Windows to install a browser separately, based on a consumer's decisions. If one answers "some" then one looks to a type of "must carry" remedy; a remedy that has been reported as of interest to the EC.

In her synthesis post, Mitchell added,

The more direct, product based Principle 3, Windows must enable people to choose other browsers, generated some very positive feedback and also some concerns. The positive response comes from the idea that one can't address the problem without addressing the product. The concerns seem based in (a) complexity of user experience concerns; (b) concern over unintended consequences.

Here's what Microsoft is offering the EC for browser choice remedies.

2. (7) Microsoft will distribute a Ballot Screen software update to users within the EEA of Windows XP, Windows Vista and Windows Client PC Operating Systems, by means of Windows Update
2. (8) The Ballot Screen will give those users who have set Internet Explorer as their default web browser an opportunity to choose whether and which competing web browser(s) to install in addition to the one(s) they already have
2. (11) The Ballot Screen will be populated with the most widely-used web browsers that run on Windows with a usage share of equal to or more than 0.5% in the EEA as measured semi-annually by a source commonly agreed between Microsoft and the European Commission
2. (12) The Ballot Screen will in a horizontal line and in an unbiased way display icons of and basic identifying information on the web browsers
2. (9) ...Microsoft may offer tools to volume license users that prevent the Ballot Screen update from being installed on all computers covered by the licence
2. (10) Nothing in the design and implementation of the Ballot Screen and the presentation of competing web browsers will express a bias for a Microsoft web browser or any other web browser or discourage the user from downloading and installing additional web browsers via the Ballot Screen and making a web browser competing with a Microsoft web browser the default.

This remedy had supporters and detractors both inside and outside of the Mozilla community. Whether you came down for or against, since Microsoft has put this on the table as one of their big offers to the EC and since it's getting the lion's share of press coverage, I'm interested in what you think Microsoft's offer would actually accomplish and whether or not it would address the principle involved, Windows must enable people to choose other browsers.

I'm of mixed feelings on this one. I think that a lot will depend on the actual implementation of the ballot. In the current incarnation, I'd be surprised if any major changes resulted.

First, by putting IE in the left-most spot, they're directly contradicting paragraph (10). That layout makes IE the "default" choice because it places it first in a horizontal list. For real unbiased ballot, the primary offerings should be in randomized order, or some other order that doesn't recreate the priorities that MS obtained unlawfully.

If the ballot is to go beyond just offering a real choice and the goal is to actually effect real change, IE should be placed last on the list so that users arrive at IE after evaluating the descriptions of the other browsers first.

IE has essentially twice the representation on this ballot screen as the other browsers, maybe more. Since it already is the default browser for Windows, "no user action" is the same as choosing IE. That means there are two ways to achieve IE as the default, either by selecting it, or by ignoring the question. That's a very large advantage for IE.

(Most common user behavior when confronted by roadblocks is to take an action they believe will most effectively remove the roadblock. We've learned this lesson well in dealing with browser security warnings and other dialogs that interrupt the user's flow. Users are going to instinctively dismiss the entire window or pick the default or most prominent choice.)

But even with all those items addressed, there's still the very large problem with this implementation. If we assume that the user choice to pick a different browser in this ballot is a clear signal of the user's intention to actually get and at least try out that alternative browser, then we have to have some better mechanism for fulfilling that intention. Clicking a download link does not guarantee that the user will achieve the goal of getting and trying a new browser. Far from it. There are quite a few steps between clicking the link and a new browser window in arriving in front of the user. At Mozilla we're very experienced with this and we probably have the most effective system in place for helping a user through the process of clicking a download link and having a first run experience with the a new browser, but we still see significant drop off at every stage of that process. For example, if you count the number of people that click the "download Firefox" button at our Website, and then you count the number of people that actually completed the download, located the downloaded file, launched the installer, and completed the installer, you'll find that only about 57% of the people make it through that process and ever even get a first look at Firefox running. And we're really really good at this. There's even further fall-off when you look at people who can still locate and use Firefox a month later resulting in a conversion of only about 35%. So, if you want to fulfill the user's intention, it's simply not enough to put a download link on the page. The effort a user must understand and put forward to achieve a browser change is significantly larger then the effort required to maintain the status quo and that just isn't a recipe for meaningful choice or real change.

Finally, even if a user does manage to acquire a new browser, IE still has the advantage of starting out with the default status and the most prominent placement on the Windows Desktop and the Start menu. For ever after, even something as simple as an accidental or habitual click on the "blue e" will result in another easy opportunity for Microsoft to undo the user's choice.

It's my view that to make the ballot meaningful, Microsoft would have to address all of these issues, and possibly a few more I haven't thought of.

The fourth principle, Microsoft's financial and other incentives to distributors must be browser-neutral, didn't generate a lot of discussion in the Mozilla community. I think this is because the OEM channel is opaque to most people, or even worse, works almost precisely opposite from how one might guess.

Here's Mitchell again:

(This [OEM] distribution channel is *so* valuable that Microsoft's early efforts to promote IE in the 1990's included threatening the OEMs with the loss of their ability to ship Windows (and thus the end of their business) if the OEMs didn't ship IE exclusively. This practice stopped after the US judicial system determined a set of these sorts of practices to be illegal.)

Historically, software vendors generated revenue on upgrades and the licensing of subsequent and additional products. Today the models are diverse and complex, and may also include revenue-sharing between OEMs and software or service providers. For example, if you use a desktop search functionality, chances are high that the company you bought the machine from is getting a piece of revenue from the search provider.

This principle does not challenge these general business models. Like the other remedies, it is tied to the monopoly status of Windows, which requires all PC OEMs to work with Microsoft. In addition these programs cannot be matched by others because the Windows monopoly gives Microsoft a raft of unique tools. This principle prohibits the use of those tools to promote IE in ways that are unavailable to other browser manufacturers. It asserts that Windows monopoly status cannot be tied to financial incentives that further damage browser competition.

Microsoft's settlement proposal says:

1. (4) Microsoft shall not retaliate against any OEM for developing, using, distributing, promoting or supporting software that competes with Microsoft web browsers, in particular by altering Microsoft's commercial relations with that OEM, or by withholding Consideration.
1. (5) Microsoft shall not enter into any agreement with an OEM that conditions the grant of any Consideration on the OEM's refraining from developing, using, distributing, promoting or supporting any software that competes with Microsoft web browsers.
1. (6) Microsoft shall not terminate a direct OEM license for Windows Client PC Operating Systems without having first given the OEM written notice of the reasons for the proposed termination and not less than thirty days' opportunity to cure.

So what do you think? Is this sufficient to calm fears that Microsoft would strong-arm PC makers into distributing IE or making it the default browser rather than some alternative?

This sounds to me a lot like what Microsoft was already forced to agree to in the U.S. after they were convicted of anti-competitive practices for exactly this kind of behavior at the turn of the century. If this wasn't already the status quo, it certainly would be a good thing to make it so.

Mozilla's fifth principle was Microsoft must educate people about other browsers.

Mitchell introduced this principle saying

One of the results of the Windows / IE integration is that millions of people believe that the "blue e" icon IS the Internet. They are unaware of of Microsoft's control over their online lives through this blue "e" or that they have additional choices. This principle asserts that Microsoft should participate in correcting the misconception that it has created. The monopoly of Windows and Microsoft in people's computer experience means there is no other entity that can substitute for Microsoft here.

And here's what Microsoft is offering that aligns with this principle.

1. (1) Microsoft will maintain a web page on www.microsoft.com that explains how users can turn Internet Explorer on and off, and will maintain that page so that other browser vendors can link to it if they wish.
2. (8) The Ballot Screen will provide two links associated with each web browser. An "install" link will connect to a vendor-managed distribution server, which, upon the user's confirmation, can directly download the installation package of the selected web browser (and only a web browser, including software to update the web browser only) for local execution (the resulting situation will therefore equal a scenario in which the user himself had downloaded and executed the installation package without being aided by the Ballot Screen). An "information" link will connect to a vendor-managed web page from which the vendor can offer users more information about its browser and installation options. Users will be able to select one or more of the web browsers offered through the Ballot Screen. Microsoft shall ensure that in the Ballot Screen users will be informed in an unbiased way that they can turn Internet Explorer off.

This is obviously not a major education campaign, but I'm interested in whether or not you all think these documents will address the principle that Microsoft must educate people about other browsers.

In my view, this principle was always going to be difficult to put into practice outside of what could be accomplished in some implementation of principle 3, but Microsoft does have a pretty big footprint in publishing and education so there's no shortage of places Microsoft could create and deploy content that educated people about Web browser choice. I think, though, that the most effective education would be delivered through Windows itself with something like the proposed ballot. The information included in the ballot would need quite a bit of improvement over what Microsoft has proposed, at a minimum prefacing the choice of browsers with a clear description of what a Web browser is and how they can differ from vendor to vendor.

The sixth principle, Microsoft tools for developing content must not produce IE specific or Windows-specific results and the seventh "potential" principle "IE must comply with web standards" (Opera has suggested that Microsoft must support web standards they have promised to support) go un-adressed in Microsoft's remedy offer to the EC but are somewhat covered in an accompanying proposals that Microsoft calls a "Public Interoperability Undertaking" which as I noted in the intro to this post isn't legally binding.

Still, Microsoft is on record saying that it will support a set of Web standards and that it will do more to increase interoperability between its products and products from other vendors.

How do you think that will work out? Do you think Microsoft will follow through with this Public Undertaking? Will it have the impact that Opera and other advocates of using legal pressures to force standards compliance were hoping for?

I think that six was actually a pretty important principle and I would have liked to see Microsoft address this specifically in its settlement proposal. Microsoft's IE hegemony has been consistently bolstered and supported by the IE-specific or IE-advantaged Web content that is produced from other Microsoft products. When you save an Office document for the Web, for example, it has always favored the IE browser and disadvantaged competing browsers. When you produce a Web site or even a simple Web page using Microsoft's Web development tools, that site or page has always worked better in IE than in competing browsers. This has the very real impact of discouraging the use of competing browsers and persisting Microsoft's dominance in the space.

The idea of legally mandating standards never appealed to me. I think I understand what Opera and other supporters of this principle were trying to achieve, but I think there were always just too many pitfalls here. It's even more than just the question of which standards and what "support" actually means. If, for example, the EC mandated that Microsoft fully support CSS standards that were ratified by the W3C. Well, Microsoft is a member organization with standing on the CSS Working Group and they could simply derail useful CSS advances in order to avoid having to comply with the EC. No one wins there. In my view, the standards bodies are already problematic enough and I really wouldn't want to see them further perverted by member organizations seeking legal or political gain.

I'll just close by re-stating that I think Mozilla's involvement in this has been quite positive and I believe that Microsoft's settlement attempt addresses many of the concerns that Mitchell spelled out. Whether it addresses them well or effectively, I'm interested in hearing from you all. I'll reserve my judgements for a later post.

Posted by asa at 10:10 AM

 

asa2008.jpg