1. It’s the most user-friendly and compatible IMAP client of the modern apps, and IMAP is gaining traction with a lot of ISPs and universities.

2. Its interface is consistent between Mac OS, Windows and Linux – a big plus for folks who use email at work and home and don’t want to deal with two different interfaces.

And the Thunderbird crew is really quick at squashing bugs, which is a big plus. My employer uses Thunderbird as its default email client (IMAP mail, Macs and PCs throughout our office), and while we support other clients (e.g. Entourage, Eudora, Outlook), Thunderbird is the best app for IMAP: consistent behavior and interface, etc.

At home, where most of my accounts are POP3, I use Mail.app.

That’s why these on-line polls are nothing more than eye candy. It’s not that they’re meaningless… They can be interesting, but at a purely abstract level. There’s nothing concrete that can be said about them.

I was using meaningless as a contradiction to any idea of them meaning more or different things than they actually do.

There is ABSOLUTELY NO THEORY that exists to explain polls taken in this manner.

The For Entertainment Purposes Only disclaimer seems well-fitting for polls with aforementioned kinds of biases and other dubious conditions.

Wrapping up this silliness by answering Tim’s original question:

Looking at these results, what surprises you?

Nothing really, about the results. Any surprise is the “serious” amount of attention such a mundane, unserious poll has generated here. :-)

Tiem and time again it has been shown that polls conducted through a volunteer method like this have little probability of representing an entire population. That is, these results have little chance of matching an actual scientific sample of macOSXHints readers let alone Mac users as a whole. The people who decide to take such a poll are not a representative sample of any population. This is exactly what happened with “Dewey WIns!” People who had telephones simply were not a good representative sample of a population. People who volunteer for polls are even worse. You skew your sampling.

That’s why these on-line polls are nothing more than eye candy. It’s not that they’re meaningless… They can be interesting, but at a purely abstract level. There’s nothing concrete that can be said about them. There is ABSOLUTELY NO THEORY that exists to explain polls taken in this manner.

Here’s an example of what I mean about stability. You can get heavy tailed/free-scale like race conditions with these polls. Let’s say that someone views your site, and you’re showing the stats half-way through the week. Someone sees that Thunderbird isn’t doing so well. They would have never read macOSXhints, but they want to stick up for Thunderbird, so they go to the site. Their vote is dependent upon the votes before them. The statistics that you are using to draw your conclusions make the assumption that each vote is indepedent of every other vote.

And so, depending on timing and initial conditions and lots of other things, if you re-ran this poll today and then next week and then next week you may get completely different distributions. You simply cannot simply assume that these are stable results. We know NOTHING about the convergence properties of polls taken this way.

So when you see a 3% or 5% difference from a poll taken last year… I don’t that shows convincing evidence of anything. I think it just does exactly what these polls are meant to do — make readers feel warm and fuzzy.

I don’t understand why people think that the figures are “meaningless”. Stastically unsound to some degree, yes. Skewed a bit, yes. But meaningless?

Why is this not an accurate breakdown of macOSXHints readers who happen by in the week of the poll, are prepared to participate in polls and willing to press a radio button?

Is there some other statistical factor that I am overlooking?

@Jeremy: I completely agree with you about Mail’s usability and smooth edges. I love it too.

Mail’s the most usable and smooth OS X MUA for me, too. But too many issues still keep me from “loving” it, especially with its potential for randomly losing messages (which fortunately I have noticed happen for awhile, though that doesn’t mean it hasn’t).

Gmail wasn’t on the first mail client poll and just its popularity has surely had a significant influence on switchers from Mail. Some switchers to Thunderbird may have been influenced by their preference for Firefox and its increased popularity, too. The second poll has almost double the number of respondents, without knowing how many participated in the original one. And the percentages seem simultaneously predictable and meaningless to me.

Mac OS X Hints visitors represent a more technically inclined and/or curious segment of the Apple/Mac community. It’s usually easy to see how that heavy bias influences polls there, which are a harmless (even silly) indicator of relatively little or no significance.

But it is interesting that lots of other people in the two polls don’t seem to share the love. I wonder what makes them switch?

One of the most important features to me is the ability to turn HTML email off. I hate HTML email (evil evil evil), and Mail.app is so far the only well-integrated GUI mail app I’ve used that actually supports turning HTML email off (via hidden preference). Thunderbird can’t do it (which is infuriating because customization is about all it has going for it). Mail.app always (and I mean ALWAYS) displays mail in my chosen monospace font and size.

I’ve been a finicky emailer for a really long time (elm, anyone?) and I think I’ve used every GUI and non-GUI email client for Unix available since 1993 or so… Mail.app is the best client on any platform that I’ve used, for what I do. I certainly wouldn’t “switch to Windows” over a mail app (as if switching operating systems would fix it, and as if Windows has any really good mail clients either)… that would be like cutting my head off because my nose hurts.