[2067] in athena10
Re: Mail folder policy
daemon@ATHENA.MIT.EDU (Evan Broder)
Sun Apr 5 09:54:11 2009
Message-ID: <49D8B7EF.8090100@mit.edu>
Date: Sun, 05 Apr 2009 09:53:51 -0400
From: Evan Broder <broder@MIT.EDU>
MIME-Version: 1.0
To: Quentin Smith <quentin@mit.edu>
CC: Greg Price <price@mit.edu>, Tim Abbott <tabbott@mit.edu>,
debathena@mit.edu
In-Reply-To: <alpine.DEB.1.10.0904050256360.7699@vinegar-pot.mit.edu>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Ok. So we can't trust network's numbers. Let's get our own.
I'm looking at bc-talk, because that strikes me as a list that is
sitting roughly in the middle of the range of student technical
competency. Also, it's the only list I'm on that doesn't have a /strong/
slant towards the technically oriented. I don't think the numbers from,
say, SIPB lists are meaningful, both because of the technical slant, and
because of the relatively small set of users.
I have a relatively hackish script in my Public -
/mit/broder/Public/detect-mua - that operates on a one-file-per-message
type of store (i.e. Maildir, which is how I store my mail). It prints
the sender's email address, then a tab, then the mail client. It also
attempts to simplify the mail client identifying string to something
relatively minimal. (Note that "Outlook" includes "Outlook Express")
If I count each e-mail once (i.e. ignore the sender field), here are the
stats I get:
remus:~ evan$ cat muas | sed -e 's/^.*\t//' | sort | uniq -c | sort -rn
967 Gmail
942 Thunderbird
618 Apple Mail
322 Webmail
175 Outlook
116 pine
48 Evolution
36 iPhone
20 Eudora
19 Claws Mail
17 Unknown
24 Other
This doesn't filter well for people who send a lot of mail to bc-talk,
though. So let's uniquify over (sender, MUA) pairs:
remus:~ evan$ sort -u muas | sed -e 's/^.*\t//' | sort | uniq -c | sort -rn
312 Gmail
188 Thunderbird
150 Webmail
112 Apple Mail
46 Outlook
37 pine
26 Evolution
18 iPhone
12 Unknown
17 Other
Like I said, I'd be interested in seeing this for other dorm lists.
Maybe bc-talk has a localized cult of Thunderbird that's unique to it.
- Evan
Quentin Smith wrote:
> I noticed this ~9 months ago when this data was presented to a focus
> group I was part of; I brought this up then, as well, and they didn't
> have an answer for why they had missed Pine.
>
> --Quentin
>
> On Sun, 5 Apr 2009, Greg Price wrote:
>
>> On Sun, Apr 05, 2009 at 01:19:48AM -0400, Quentin Smith wrote:
>>> The statistics you reference were collected from outgoing.mit.edu based
>>> on the X-Sender header in each outgoing message. Since pine/alpine
>>> doesn't send an X-Sender header that identifies it, those statistics
>>> did
>>> not actually capture the complete set of mail clients.
>>>
>>> (i.e. the numbers are wrong.)
>>
>> Thanks for noticing this! On zephyr Evan reported the least common
>> client as having about 150 users, and it didn't seem plausible that
>> Pine was less than that.
>>
>> If someone wants to try to get a sense of the real numbers, the
>> message IDs identify pine/alpine clearly and usually other clients.
>> We don't have outgoing.mit.edu logs on hand, but a big discussion list
>> will do for a rough estimate.
>>
>>
>> When I've done this at Harvard a few years ago, the answer was, very
>> roughly, equal parts webmail, Pine, and Apple Mail+Thunderbird.
>>
>> Greg
>>
