- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Tue, 16 Jan 2007 15:04:29 +0000
- To: Phillip Lord <phillip.lord@newcastle.ac.uk>
- Cc: public-owl-dev@w3.org
On 16 Jan 2007, at 14:46, Phillip Lord wrote: >>>>>> "BP" == Bijan Parsia <bparsia@cs.man.ac.uk> writes: [snip] > BP> Usability is more *difficult* to measure than worst case > BP> complexity because it is highly multidimensional, and typically > BP> requires experiments. Usability is sensitive to, among many > BP> other things: task, user, and support environment (tools, books, > BP> accessible gurus). Even when one has good data it can be > BP> difficult to interpret and even more difficult to generalize. > > Well, I tend to think of "measurability" as being an boolean value. My > point was that you can measure usability, as you can anything > else. Might be hard, but it is doable. I don't really want to get into a huge debate about measurability theory, but "doable" in this case doesn't seem to mean "feasible". I think it's fair to interpret "we don't have metrics" as "we don't know how to measure these things in a way at a cost that is clearly helpful". Part of measuring is deciding *what* to measure, and in usability there's a lot of *different* things you can measure. Some, I warrent, are incommenserable, thus I would be rather skeptical about a single usability metric. > BP> BTW, I don't think any "theorists", at least, any one involved > BP> in these debates, conflate worse case complexity with > BP> effectively implementability. > > I wasn't trying to imply that they did. My point was that measuring > usability is often ad hoc, non-generalisable. We agree. > But, then, so does > tractability, Formal worse case tractability is neither, I think, ad hoc nor non- generalizable. > or effective implementability, or how fast it goes. [snip] These are definitely subject to ad hoc-ness and non-generalizability. And I agree that this is what we should care about. Formal complexity analysis can give us clues there. The TF document does not *only* inlcude formal complexity analysis, but presents languages for which there is reasonable evidence of other virtues including useful expressivity and reasonable implementability. I think we are on the same page. Cheers, Bijan.
Received on Tuesday, 16 January 2007 15:04:14 UTC