Tuesday, March 18, 2008

Interface design

Oh, sweet, sweet procrastination. I have defragged my hard drive, run a virus scan, reindexed all my files, but there is always you, Internet. You are my favorite source of procrastination. Uh, (cough) sorry.

(Edit: While typing this, you might think I just had an idea shot down or a frustrating meeting. that's not the case. The articles linked toward the end were the thing that inspired me to write this.)

When designing tasks, especially for human types, it makes sense to take a step back and make sure you covered the basics. Too many times I see task design go to committee when all that does is soak up time and energy. There are two parts to any interface design: Style and Technical. Style is that inexplicable, non-critical bias we all have. Technical ensures that the task answers the intended question. This all goes awry when we start nitpicking. By nitpicking, I mean minor changes based on loose evidence that happens to be significant in some cases, but not in the one in question. I strongly, strongly, strongly urge all labs involved in interface design for behavioral tasks to do two very simple things:
1) Limited the number of people designing the task to 3 at the maximum.
2) Make a checklist that every task must pass.

Three people? Yes, well no. Two people! The process goes like this:
START with style. The 'creative' person (who also implements the task) should design the interface. The 'technical' person should ensure that it answers the question. I'm assuming a lab situation where no one is an idiot, so the technical person is really scrutinizing the actual science and comparing it to other similar tasks with a more critical eye wile the creative person is more concerned with fitting it into their overall scheme/study. The creative person should be the less experienced, the technical a post-doc or PI. Any concerns pointed out by the technical should be handled by the creative in whatever way they see fit. The result should be discussed between the two and then presented to the third, who has had no influence on the design. This meeting should be no longer than 30 minutes.

So, it goes: Creative -> Technical -> Creative -> Technical/Creative -> Outside Confirmation (-> back to Creative/Technical if significant issues identified)

You may say, "That's what we do!" I doubt it. Having worked in five labs and three companies, no one has an actual process. It generally begins with a discussion in a group where personal bias is introduced and emphasized before the actual task or interface creation. Advice is sought by multiple people, sometimes because of expert opinions or a wish to collaborate with a specific person, which means the nice linear design above branches by N! times as each change and objection is propagated throughout all involved members.

It is still important to get everyone's feedback and make sure other lab mates know about the tasks, so presenting it to others is still a good idea (isn't that partially what lab meetings are for?), but when presenting, do the following:
- Any idea that doesn't relate or has previously been discussed between the group of three should receive a, "that's a good idea, I'll look into it" response. Do not explain, do not argue, and at most ask why that's significant. You do not need to rehash the last 4 months of interface design to someone that wasn't involved. They were interested, they had a good point, thank them, maybe rehash in your head what your previous conversations were about concerning it, and move on.
- Any undiscussed ideas or unknown sources ("The X lab did something like that, but did X(i)") should be written down and assigned a value 1 to 3. 1-not important, 2-medium or UNKNOWN importance, and 3-Crap, should have thought of that. Address those issues in that order. Address the #3 issues immediately after the meeting and look up the info for unknown sources. make sure all three people have assigned the same values to the identified issues and leave the creative/subordinate to iron them out. Repeat the linear approval process above once these new issues have been resolved.

You are now done. Sounds long, but it really is about 1 week. Each step should take around half a day so you have time to mull over the info. Don't throw things up the line that aren't done, don't meet in between to discuss anything other than the vital issues and any possible implementation problems. The creative person is the one most vested in seeing the task done right, and will make sure the design goes through.

This is how businesses work when they hire designers. The designer(s) (usually 1-3 people), meet with the client, find out who they are trying to attract, get enough technical details to know what the business is and is not, leave and take time without any company involvement until the next formal meeting, meet to present several sketch ups of the final product, return to the studio to alter the chosen mock up to the specifications, call/email/fax the final design, get the approval, and make the widget. For bigger things, like whole websites with many categories, each meeting will discuss a specific subcategory after an initial unifying design meeting (similar to a student meeting with the PI to discuss his/her thesis work).

In order for this linear design to work, each hand off of the task should be preceded by a checklist.
A) Who else has done this task? What were the major problems and are they more or less a concern for you?
B) It is technically and financially feasible?
C) What about the currently available setup needs to be changed?
D) Does it conform to the following (what inspired me to write this whole babbling article):

LAST) Does it still answer the questions you are looking to ask?

Alright. Back to work.

No comments:

Post a Comment