Our success in the market got others interested. With more people in the market, the price eroded. In three years, we saw the cost of a sensor go down by a factor a four. The key point is that we got to learn the market when there was a lot of profit. Those coming into the market a few years late have to make the same mistakes we did, but they don't get to profit, so to speak, as they learn from their mistakes. At some point, it will likely no longer be cost-effective to try to get into the capacitive touch sensing business.
Furthermore, the solutions we provide today look nothing like our original programmable designs. Each new product builds upon the knowledge base developed on previous products. As a result, we have must less to learn with each design cycle.
I once had a boss who thought the power brick business looked good and asked me to produce a reference design using our part. I went to several power brick companies and found out just how much I didn't know.
I came back and told my boss that I could design an 85 percent efficient power brick that would get us laughed out of customer meetings. He implied it should be easy, and I showed him a power brick schematic. I told him that I had found five different feedback paths and had figured out what only three of them did. I also suspected that there were some exotic magnetic feedback paths.
I told him it would take me three years to come up with a competitive solution, and even then, at best, the solution would still be three years behind the competition, which would have already moved on. Those power companies started the market when an 85 percent solution was acceptable, and they had been making small improvements one product at a time.
The value of the common knowledge base
Another company I worked for decided to shut down a product line and fire all the senior (read "expensive") staff. A few years later, it decided that this had been a bad decision, and it wanted to get back into the market. It found it could not do so with the documentation it had, and none of the staff that had been let go wanted to come back.
A key point of Kurt Gφdel's work is that it is not possible to design a system complicated enough to describe itself. Within any system, there is a core of information that lives within the culture of that group. Break up the group, and that information is lost.
This is good news for engineers involved with companies that are successful in a market. For example, many companies decide it would be cheaper to acquire a successful group as a shortcut to developing the core of information on their own. (Look what TI paid to acquire Burr Brown.) Alternatively, a company may decide to hire away members of a successful team to get a start on that core. Companies will offer you money, promotions, and stock options to obtain the experience you have painfully acquired. Even if you decide to stay with your current company, the company understands what a dear resource you are and may make financial adjustments.
When getting into a new market, time to market is essential. It is important that you make the mistakes you are going to make while there is profit to pay for it and lower expectations on product performance.
Dave Van Ess is principal applications engineer for Cypress MicroSystems.