Re: list o' importants

[Roger Dingledine <arma@mit.edu>]

In message <35A4E947.CF82A1B2@ix.netcom.com>, kmself@ix.netcom.com writes:
>> response later and peruse or modify it. And I was thinking about keeping the
>> email address around in case we needed to contact the person.
>  
>Email is a reasonable proxy for a personal identifier (not perfect, not
>terribly bad).  It also allows validation of recipient (send a
>confirmation message to the address -- if it's bounced, the address is
>no good, a response is not required).  Other validation methods are
>possible.  Slashdot checks IP addresses.  I don't think this is very
>effective, given the number of people using ISPs with dynamic IP, and
>possibly limiting, given firewalls (see the Slashdot survey page at
>http://slashdot.org/).
 
Just poked around slashdot a bit for a page describing their survey
technique, and I didn't find anything other than a pile of silly questions
and a larger pile of inane comments. Slashdot should be burned at the stake.
(Oops. Did I say that?)

>I think you're thinking too hard.  I see the following categories,
>basically an extension of the Core/Layers concept, with divisions for
>Personal, Workstation, Server, and Portable categories of use.  I get
>seven categories, two or three of which may not be of particular
>interest.

Excellent idea! I shall include a few rules of thumb here, for our
discussions on this topic:

a) We want to have as few categories as we can get away with. That
   way we have a lot of data points in each category.
b) In particular, we want to avoid splitting up categories in which
   we're unlikely to get very many people anyway.
d) Unless we're planning to do fancy statistics, which I don't think we
   are, the order and grouping of questions is not going to be
   statistically significant. However, we should seriously consider
   having *different wordings* of each issue based on which user type
   we're asking. That will make it more difficult to compare between
   user type, but it will allow us to get at the heart of the issue
   that we think matters most to that user type.
e) Except we shouldn't pre-create assumptions about user types.
   Specifically, we need to throw away statements like "Ease of admin
   or availability of technical service support important", because
   when we get around to actually picking which questions go to which
   user type, we will end up picking predominantly "ease of admin" or
   "tech support" questions for that user type, and we will skew our
   results. The point is that we're trying to figure out if it's
   important, not prove that it is. (This is what a survey-clueful
   person just told me, at least. My instinct says we can keep it in
   mind and still come up with a fair survey. There's probably a good
   compromise somewhere.)

But anyway, now I'm going to go through and nit-pick.

>Categories and inclusives:
>
> - General Home user

I don't like the word 'home' here, because it confuses students and
other people who would otherwise fit in this category. Something like
"General (Home) user" might work.

>   Gamer, Internet browser/email, wordprocessing, schoolwork, personal
>accounting,
>   other light tasks.  Possible (but unlikely?) local network.
>
> - SOHO -- Small/Home Office
>   Business applications (wordprocessing, spreadsheet, presentations,
>database), 
>   fax, email, web, small local network likely, possibly specialty
>software.  
>   I'd include graphic artists,  musicians, and other artistic types
>here.  
>   Ease of admin or availability of technical service support important.

My boss in high school was a scientist; he was interested in having a
stable platform from which he could run his custom-built data-sampling
programs. It was based on custom hardware, and he didn't use spreadsheets,
word processors, accounting, games, etc. He just wanted a machine that
worked for his specific purpose.

We should consider splitting the "graphic artists & co" from the SOHO
category and merging them with the scientists in a "specialty user"
category, characterized by custom hardware or software and a very
limited and nonstandard range of activities performed on the computer.

> - Development/Technical Workstation
>   Heavy duty technical use, programming, hacking, applications
>development, etc.
>   Local network likely.  Very technical user.  Specialized software. 
>Not sure 
>   this fits the SEUL target model, though as a technical WS, it might.

This will be by far the most frequently used category in our first phase
of surveying, given our initial audience.

> - Business User -- Mid/Large office
>   Like SOHO, but more so.  Significant networking -- file, print,
>application
>   serving.  Probable use (but not admin of) servers, firewalls, etc. 
>Personal
>   business applications (wp/spreadsheet/presentation/database). 
>Enterprise
>   applications.  Ease of use, GUI are issues.
>   Probably the user we care least about because they don't make
>   purchasing decisions.
>
> - Business User Systems Admin -- Mid/Large office
>   Use categories as above, but this is aimed at the folks who
>administer the 
>   stuff.  More emphasis on capability:cost ratio, ease of admin. 
>Again, not 
>   sure that SEUL points this way, though many of the questions we've
>been 
>   coming up with fit this category best.
>
> - Server Class
>   Applications, file, network, Internet, mail servers.  Also possibly
>embedded
>   or dedicated systems.  Aimed at admins or technical types.  Not sure
>this
>   fits SEUL target audience.

This may well be a separate category from the rest, but do we really
care about them enough to have a separate category for them? Are there
related categories that they might fit under almost as well?

> - Portable
>   This is essentially a hardware repackaging of SOHO/Business user.
>   Think laptop, Itsy, handheld.  Keys:  applications, power saving
>features, 
>   speed, compatibility, networking, communications, and data transfer.
>   Ease of use, ease of admin, GUI, as for SOHO/Business user.
   
We should all think of each of our relatives and friends, and try to
force them into one of the above categories. If you can't make them
fit, tell us. :)

Now, the next problem I have is...these are "what does this person
use a computer for?" questions, not "what is this person's computer
used for?" That is, I may have one setup at work and an entirely
different setup at home, and I have different requirements for each
and they fit under different models. For instance, somebody might well
be a "business user systems admin" at work and a "development/technical
workstation" at home. Do we just tell them to pick one? Do we let them
fill out the survey multiple times, one for each? (If the latter, I
definitely want to keep track of who is who in terms of surveyees.)

--Roger