So what is going on here? Are engineers simply irrational about IP? Not really.

Knowing that IP is necessary in order to meet schedules, engineers find it. But knowing that integrating the IP will not be a pushbutton process, they search at least as much for a vendor that will work with them on integration and verification as for exactly the right piece of IP. In fact, they may ask a trusted vendor to develop a block it doesn't currently offer.

The centrality of vendor relationships was one of a number of interesting points that emerged from our study of intellectual-property selection habits. Another was simply the pervasiveness of the problem-solving ethos in which engineers are trained. In our focus groups, many engineers said their approach to IP selection was quite ad hoc. Yet in the large-sample study, we learned that even those who thought they were having an easy time of it followed essentially the same process as those who considered the problem difficult and approached it formally.

In fact, what makes the same process hard for one group and easy for another appears to be circumstances that let the design team shorten steps in the process. Respondents who said their selection process was easy credited the availability of silicon-proven IP, a known fit into their existing design flow and an existing relationship with the IP vendor. There is no shortcut to the painful process of reducing the risk of IP acquisition, but there are circumstances that can abbreviate some of the steps.

The point about existing vendors raises another big issue — the business model problem. Our survey showed that a significant amount of the total time required for IP selection goes into negotiating business issues. There is some evidence to suggest that this is the task in the selection process that is the least predictable, as well. Moreover, at this point in the process, nontechnical management becomes involved to a greater degree than in any other engineering decision studied. Certainly, corporate managers wouldn't be around if the perception of great risk did not exist.

One-time model preferred
When we asked engineers if price was an issue in IP selection, they said two things: A low price can make the choice easier, and a high price — or at least a high enough price to get the attention of management — can make it a great deal harder. Asked for their preference in license models, a considerable majority want a one-time purchase model, not clever, limited pay-to-play models that spread out the payments but complicate the process. Of course, it is possible that we would get a different response if we were to survey purchasing managers or CFOs.

But the degree of emphasis on the negotiation phase, on the part of the whole organization, clearly says that the business model is broken. One hears of negotiations that simply fail or that, if examined for the time consumed, cost the vendor and the user together more than the value of the contract. Certainly, the negotiation process intrudes into the technical decision making and may even lead to suboptimal choices of IP. Given the costs involved in the chip design process, it seems imprudent to let a small difference over license fees, rights or payment models bring about a large increase in project schedule risk. But not infrequently, that's exactly what appears to happen.

It's tempting to suggest that the industry needs a set of standard license templates and an open, standardized pricing model. In the study, engineers were quite clear on what they considered to be the value of IP: It potentially reduces project cost and risk. It should not be rocket science to quantify those benefits and use them as a basis for pricing.