I'm actually torn on the issue. I can see that a patent has the advantage of conferring exclusivity, but time and money consumed might increase time to market and the "sunk cost" of a patent and might deter a much needed change of direction or technology.
I've been touched by patents twice. The first time was as an employee of a start-up based on a patent by a PhD student. In that case I think the company got blinkered by need to "implement the patent" and got overtaken by others who were more flexible.
The other was commercialisation of WiFi. CSIRO owned the patent. Radiata the company did not own any patents (not sure if they licensed the CSIRO one). Within 3 years, Radiata got acquired by Cisco for AUD 560 million. I was later told by one of the founders that Cisco had decided to acquire whoever was first to market, and Radiata got lucky and beat Atheros by 2 weeks. Investing 2 weeks into taking out patents could have been very costly!
CSIRO eventually made about AUD 1 billion off their patent, but it took 20 years, involving a 10 year legal battle that cost them AUD 15 million per year.
Since then, my feelings have been that patents are fine if you are well established with deep pockets, but a potential liability otherwise. I'm always interested in data that challenges that view as I could well be wrong!
It's based on work I did for my Honours project. I wrote a bunch of prototype code and a pretty detailed description of its design and operation.
But in actual fact the invention is a miniscule part of the work that has to be done to bring it to market.
If spending a few thousand dollars now can help me raise hundreds of thousands of dollars later, I view that as a smart gamble to take.