I should say that in general we do need knowledge developed in public institutions including unversities and government labs to have a positive impact on society. What concerns me is that if the wrong conceptual models are applied and with them the wrong metrics then we could easily have a perverse effect.
To avoid perverse effect we typically need to map larger sytems and their interactions than we are used to.
So let us start with the example of the national accounts of economies - we know that rapid economic growth can be created by simply strip mining all the natural assets and then selling them on the world market. A case of this might be the case of phosphate mining on Nauru http://en.wikipedia.org/wiki/Phosphate_mining_in_Nauru . Such an approach is obviously crazy in the long term. Short term cash traded off for long term unsustainability is non-sense. In response to the overarching criticism of GDP the Australian Bureau of Statistics (amongst a number of national statistical agencies) have begun working on net wealth indicators and has now expanded that even further to the measures of progress project. The point with net wealth indicators is that if you remove mineral resources now, (in a accrual accounting rather than cash accounting system) it is the eqvalent of selling assets.
So how can we apply this to knowledge which obviously has different properties to minerals in the ground as it is harder to trap and sell as it is there for re-use later. Commercialisation policy has exactly this goal to quarantine knowledge into saleable units for pushing into the market. Thus it does have certain dimensions where pushed to extreme the comparison is valid. The metrics look better the more that is sold, and that is achieved by packaging parcels of knowledge.
Now at this point I need to introduce another concept - reserve deposits. In mining as we exploit the easier to access resources we are driven to harder and harder locations (deep sea oil drilling and tar sands might be examples). The more we demand, the more the price goes up and the more rapidly we exploit the available resources, thus creating an incentive for accessing harder to obtain resources. So we end up and image like this one. The more we access more costly resources the more the price will go up.
Fig 1. We can illustrate this process.
In contrast to this what we see with Universities and government research is a different process. Government tend to want to create incentives for as much activity as possible to be commercialised. To achieve this it is necessary to set the selling price at a point where there will be buyers. In a competitive environment and knowing that university knowledge is often very expensive to develop to point where a business can make money then the only way to sell the knowledge is for the price to be driven down.
The only way this can happen is if the packets of knowledge get smaller and the projects thus less innovative. Unfortunately, what we know about kowledge is that the more complex the problem the more we need to draw up a rich base of knowledge from multiple fields.
In general this falling quality is exactly what the OECD has found The Science, Technology and Industry Scoreboard 2011 finds that patent quality has declined by an average of around 20 per cent between the 1990s and 2000s, a pattern seen in nearly all countries studied.
If we don't develop some metrics of net value of research and that isn't just costs and sales - that is still cash accounting but a sense of the assets and liabilities in terms of knowledge pools we will continue to have no real idea about the effects of science policy and it will continue to bouce around as it always does.
