Sat May 16 10:57:36 EDT 2009
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
The prophets [Jakob],[Joel] and [Tog] proclaim good programmers watch
users using their software and bad programmers assume they
know what users want. Swetha Chandrashekar and I spent
Wednesday watching users.
Our big goal was to find if the [new] version of the site
was better than the [old] version of the site. We
visited a clubhouse for people with [mental] illness and
tested our assumptions.
For example, the prophet Jakob says people usually read
only the first 11 characters [11 char] of a link. We took the
links from the old and new sites, put them on slips of
paper and showed them to users one at a time.
[11 char] http://www.useit.com/alertbox/nanocontent.html
The simple act of making up the surveys pointed to some
likely problems. We didn't expect many people to guess
the purpose of:
Get Your Age
We were surprised by links that we thought were fine but weren't.
... were not ok.
People confused contacting web site administration with
contacting the agencies in the database. On reflection,
reality matches our research. Once a month or so we'll get an
"When does your nurse training program start?"
"How much are the raffle tickets?"
Next week we'll see how we do with
Our next test was to give people a piece of paper with a
bunch of links, including some links from sites that had
nothing to do with ours.
Get Your Age
We asked people things like:
"If you worked for a non-profit group, what
would you click to add your group's
info to MVHub?"
The popular answer was not (as I assumed, it would be):
Putting aside the paper, We gave our subjects a list of
tasks to solve using the actual websites. For example:
"Get a phone number for the Community Software Lab"
"Who is the person in charge of CTI"
"Help a homeless person."
"Find info about dancing lessons"
Almost everyone failed at searching for a specific
agency. We have two search boxes "Program" and "Agency".
Typing "CTI" into the "Program" search didn't work.
Our most recent test subjects had very low literacy. (4th
grade?). Even if illiterate people weren't part of our
target audience, the results still would be useful. Our
previous subjects, (people with graduate degrees in
computer science volunteering with us), were served
almost as badly by our website. Both groups used the
"wrong" search box as their first effort.
The difference between the two institutionalized groups
was that the CS grads tried other strategies (reading the
on screen instructions), when their first quick try
failed. In real life, I don't know that the either group
bothers with a second try.
Sometimes different groups need different interfaces.
Linux system administrators grow impatient with clicking
"next". Other people don't see '--recursive' as self
However 95% of the our users are going to be infrequent
and impatient. Curb cuts help marathon runners as much as
they do people in wheel chairs. We need to simplify, simplify,
simplify and test, test, test.
Of course, our excitement at the useful work ahead is
tempered by pity for those less fortunate than us: