A study carried out at MIT seems to indicate the possibility to better forecast technology evolution by looking at patents.
Mind you, it is not a sure thing, but the analyses done and their application to the past seems to provide interesting confirmation that indeed by mining patent information you can get some robust prediction about the future, at least in terms of trend.
The work has taken into account 28 technology areas, from genomics to 3D printing, from photovoltaic to optical switching. The team has dug into over 500,000 patents all over the world. The first stumbling block has be to create a uniform view of these patents. Patents clustered under optoelectronics, material science, electricity generation might indeed all refer to application in photovoltaic area. Hence the need to understand what they are really related to to make meaningful analyses.
The team developed a number of algorithms and tested their sensibility based on the available historical data.
In their report they indicate that these algorithm can be used "with a grain of salt" to help in forecasting the evolution of technology.
Personally, I think that there is a need to complement this data set with information on market evolution to take into account market influence.
At IEEE we are considering digging on the vast repository of data that have been, and are, accumulating in terms of papers and presentations at conferences to provide intelligence on emerging trends.