I accept that working memory may not be clearly understood (at least by us
) but if it is
what you can hold in your head while you work on a problem or something similar (regardless of the model) I don't see it as something that can be compensated for completely by a good long-term memory or by use of external aids. Sure, one can become a problem domain expert by salting away a lot of knowledge long-term so that it's easy to recall; but still, one's thoughts about the pieces of that long-term knowledge that are relevant to the current problem are going to be created and manipulated in working memory. The working memory capacity is a limit on the number of co-thinkable things.
For me this means that for certain types of problems, which involve lots of parts that are best considered together, working memory is still important. Could one approach the same results by writing and rewriting things on paper, keeping them in front of one's face while working on the problem, etc.? Sometimes, but I don't think all the time, due to the pushing-in/pushing-out problem. Problems which can be done just as well with external aids would be the ones where the maximum number of things that must be considered at once all fit into working memory; the external aids would just preserve the state of one's work on the problem and its subparts until it was revisited.
