 |
 |
Changing subject line: too many articles in the earlier thread I do not like speaking of error or failure. I prefer to speak of learning. As I like to say, Scientists don't say the have "failed." Instead, they say "that didn't work Let me try something different." In the case of the Goal-oriented, constraint based system I described (Autodesk's Dreamcatcher), the system never makes what anyone would call an error because it should be thought of as making suggestions. The designer passes judgements, encouraging some directions, discouraging others. Think of it as critiquing. In that sense, the system is learning from its "errors". That is, it soon learns which directions to avoid and which to pursue. (It uses a learning method called "genetic algorithm": basically, it uses the machine version of sexual reproduction, sharing parts of its code with others that produced good results, shunning the code of others that have given poor results. Sex, but without any pleasure. ===== You (David) said: "Could we teach a machine to recognise the value of errors and mistakes? I (and my colleague of several decades ago, Jim Reason) argued 30+ years ago that there were two kinds of errors: Slips and mistakes. So it is not appropriate to speak of "errors and mistakes." A mistake is a special kind of error/ A slip is when the action performed is not what was intended. A mistake is when the intention itself is wrong: mistakes are far more serious and more difficult to detect. Don On Mon, Nov 14, 2016 at 6:03 PM, David Sless <[log in to unmask]> wrote: > Making errors and mistakes is important. I know we sometimes think of > errors as examples of dysfunction. I think, particularly in area like > design or poetry or painting, errors and mistakes are essential as part of > the process of making things. They are productive moments. > > The trajectory of most software design is towards eliminating errors > (bugs). Could we teach a machine to recognise the value of errors and > mistakes? > Don Norman Prof. and Director, DesignLab, UC San Diego [log in to unmask] designlab.ucsd.edu/ www.jnd.org <http://www.jnd.org/> Multiple faculty positions in design at UC San Diego http://d.ucsd.edu/jobs/ ----------------------------------------------------------------- PhD-Design mailing list <[log in to unmask]> Discussion of PhD studies and related research in Design Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design -----------------------------------------------------------------