tyler butler

Outlook, Email, and CSS

Update July 13, 2009: Thanks to some comments, I’ve got new links and minor updates in the post now.

Note: I work for Microsoft, in the Office division, but I don’t work in or with the Outlook team. I don’t have any specific knowledge about their decisions or plans, and this post is based only my own experience here at Microsoft.

The web community has been up in arms the last day or so about a campaign via Twitter pushing for Outlook 2010 to stop using the Word rendering engine for HTML email. I engaged in a short friendly discussion with my buddy Vlad on Twitter about the issue, and that got me thinking about the issue a bit more deeply. And the more I thought about it, and read everything that was being written, the more I realized that people aren’t actually communicating effectively, and that pisses me off…

The fixoutlook.org website says this: “Microsoft has confirmed they plan on using the Word rendering engine to display HTML emails in Outlook 2010.” (emphasis added) Based on that, and the rest of the text on the site, it seems like the big beef they have is that Word’s rendering engine for HTML is not up to snuff. Fair point – it isn’t. But they contradict themselves in their updated response to Microsoft’s response when they say, “We’re asking that the HTML produced by the Word engine be standards compliant. This in turn will ensure that the engine will correctly render standards- based emails.” (emphasis added)

Wait a second. Do they want an editor that produces HTML, or a rendering engine that works properly? Doing the work to make sure the editor is producing clean markup might produce a rendering engine that renders nicely, perhaps it even should produce one, but that doesn’t mean it will. Those two pieces of work are not the same thing.

Lots of people have been pointing to a post from Zeldman about this. Zeldman’s a brilliant guy, but he completely misses the point when he perpetuates a rumor about why Outlook chose Word’s engine: “Rumor has it that Microsoft chose the Word rendering engine because its Outlook division “couldn’t afford” to pay its browser division for IE8. And by “couldn’t afford” I don’t mean Microsoft has no money; I mean someone at this fabulously wealthy corporation must have neglected to budget for an internal cost.”

Ummmm… no. In my experience, the phrase “couldn’t afford” at Microsoft R&D has nothing to do with monetary cost. It has to do with engineering cost. Features don’t exist just because you want them to exist. (An aside: Raymond Chen has a great quote about this that I can’t find. If you have a link, please let me know.) So I imagine that the rationale in the Outlook team went something more like this:

Well, we want Outlook to produce really rich emails, and we want it to have a really familiar look and feel, so people already using our products will feel at home. Hmmm, building an editor is extremely hard to get right, plus we, being part of the Office suite, have several editing tools already. Emails tend to be a lot like documents, so Word seems to be a reasonable choice. This way we can leverage all the editor features Word already has, plus things they’re building this release, and focus on the Outlook-specific work we have to do. We don’t want to invest our engineering time in building an editor when we already have one.

From this standpoint, it’s easy to see why using Word for rendering as well is the next logical step. You could argue that they should use IE (or Gecko, or Webkit, as Zeldman does, but again, just because those engines are free doesn’t mean using them is without cost) to render email, and Word to author it. That’s a reasonable idea. In fact, the Outlook team seems to agree with you, because if you get an HTML email that looks wrong, you see a link to open it in a browser. In fact, even the example image at fixoutlook.org has the “info bar” at the top that does this.

You can take issue with the implementation, certainly, because it sucks mightily. The fact that I have to open the email in a browser separately blows. I shouldn’t have to do that. But don’t kid yourself into thinking that integrating an IE window into the message window is so stupidly simple that Microsoft is maliciously avoiding doing it to somehow screw users. It might be straightforward – I honestly don’t know. But do you? If you think you do, please go read Steve Yegge’s post “Have you ever legalized marijuana?” and come back.

What really bugs me about this whole thing is that people immediately jumped on the fact that Outlook uses Word for rendering, instead of sticking to the real problem that Outlook’s rendering of HTML sucks in some cases. The rendering engine they’re using is immaterial, really, because if the team goes and fixes the rendering inadequacies, then the issue goes away. They can choose how to fix the issues, should they choose to fix them at all, but at this point, even if all the rendering issues are fixed, many people will still be pissed because Word is used to render the email. The argument has shifted from being about the support proper email display to being about Word used for rendering, so there doesn’t seem to be a path to redemption for Microsoft in the court of public opinion that doesn’t involve ripping out Word completely.

I completely agree, personally, that web standards serve as a reasonable basis for email format standards even though there is no formal effort to standardize email. But please argue about the right things. Spend your energy trying to see those standards acknowledged rather than perpetuating this silly argument about ripping out Word from Outlook. Hopefully this effort will have the desired effect, and these rendering issues will get resolved prior to Outlook 2010 RTM. But I can almost completely guarantee that if you want Word completely ripped from Outlook, you’re not going to get what you want.

By the way, does anyone have a screenshot of what the email used in the example image looks like in Outlook 2007, which also used Word for rendering? That struck me as a strange omission. I’m wondering if the issues displayed in that screenshot are just bugs in the 2010 beta. I doubt it, but would still like to see it.

Update: Turns out you can see this in the Email Standards project’s review of Outlook 2007. As I suspected, no real differences.

Also – is there any way to get a sample HTML email from the Email Standards Project’s email acid test? Seems ridiculous there’s no “email me this sample email” form on the acid test page. I can and will file a bug against Outlook if I can get a copy mailed to me.

**Update: **You can now mail yourself a copy of the email directly from the acid test page.


This didn’t really fit into the overall flow of the post above, but I still think it’s reasonable info to consider, so I’m including it here anyway. I’m about to throw out a bunch of numbers and statistics that are not backed up by any data. They are based only on my own logic and occasionally rational mind. I think they’re true and reasonable statements, but I welcome data that contradicts them.

I think it’s safe to say that a majority, say, 80%, of Outlook users use it with Exchange. Also, a majority of Exchange users use Outlook, and Exchange is primarily used in business settings. Since a large majority of email sent in a business setting is sent to other people in your business, then they’re probably also using Exchange, and also probably using Outlook. Based on this not-so-scientific reasoning, I argue that the number of emails received in Outlook that didn’t originate in Outlook is relatively small. That means, practically speaking, that as long as Outlook can render email that started in Outlook, you’re hitting the majority of your users’ needs.

Now, the idealist in you (and me, for the record) is screaming bloody murder, because you want to see the “right thing” happen for all cases, not just the majority case. But unfortunately, software is more about practicality than idealism, and at some point some smart, but possibly naive people in Outlook made a tradeoff. I’d say with 99% certainty that at some point a developer or two in Outlook estimated the cost of different approaches and implementations, and this one wound up cheaper. They made a cut. They made a tradeoff. And we disagree with the tradeoff.