The two most important things I’ve worked on in the past seven years are Chebfun, a software package that people are using around the world, and Approximation Theory and Approximation Practice, a bestselling SIAM book that is changing perceptions of the mathematical properties of polynomials. It’s a safe bet that among my “outputs” of the past seven years, these two will have the greatest impact.
Not according to the UK’s assessment scheme known as the Research Excellence Framework, however. Our department has just sent in its REF submission, reporting for each of 147 researchers their four best outputs from the past seven years. In theory, the REF welcomes nonstandard outputs, including software. In practice, everybody is scared to death of deviating from the safe pattern of specialist papers in specialist journals. Luckily I’ve produced enough of those too.
So neither Chebfun nor ATAP has been included in our REF submission. Among the more than 500 outputs of Oxford Mathematics for the period 2007-2013, they do not appear.
[13 December 2013]
trefethendotnet 
I would hope that when someone with your experience submits “nonstandard” outputs for such evaluation, then these forms would quickly become standard. Perhaps we can hope.
I’d say this is the price to pay for whatever amount of academic freedom we have. For in the name of academic freedom we should receive mean of sustenance regardless of usefulness of our output to people outside academia. So we go on to just evaluate the amount and “quality” of output within it.