Every year major companies such as\u00a0Cisco,\u00a0IBM,\u00a0Lucent\u00a0and\u00a0Nortel\u00a0spend 6% to 10% of their budgets on research and development, yet are often blindsided by hot young companies. Last month,\u00a0Juniper paid $3.3 billion for NetScreen\u00a0- and I couldn't be happier because Battery Ventures, a firm I co-founded, had NetScreen stock. But it raises the question of why the industry's large R&D laboratories never seem to get it right.Imagine you are a CEO of one of the major communications equipment companies. Publicly, you are a big fan of your R&D department - and privately, you wonder what the hell is wrong. Your products are always late and seem enfeebled. You would prefer to grow organically through your own research than to be constantly buying companies and overpaying. The investment bankers camp on your doorstep with the "miracle" new company that can get you back in the game. You are in good company: JDS Uniface paid $20 billion for E-Tek; Lucent spent $20 billion for Ascend; and Nortel paid $5.5 billion for Alteon, $3.3 billion for Qtera and about the same amount for Bay Networks.Maybe the R&D departments of the industry are just plain obsolete. There are three reasons why this is the case. First, they tend to spend 80% of their time tweaking existing products and not building the insanely great new products that their customers want. Second, they are loath to come out with better products while their cash cows are still churning out predictable profits.The third reason is venture capitalists have the ability to entice the hot teams right out of the labs and build the product these equipment companies should have - faster and better. Bell Labs was a great institution but probably couldn't be put together today with the crack engineering teams that made it famous and feared by competitors. Nor could Xerox PARC or IBM Labs.The CEOs in the industry know their companies and their jobs depend on getting R&D right - and they have tried everything. Outsource R&D or bring it back in-house? Invest in venture capital funds or suck up to major universities such as Duke and Massachusetts Institute of Technology? Acquire tech companies or just make strategic investment? Whatever they do, it just doesn't seem to work.An interesting twist on this is what Cisco is trying now. Cisco took an interesting internal development, staffed it with some of its best people, gave it some money - and set it free, with a "buyback" provision built in at a pre-arranged price when the new company hits its benchmarks. Very creative and maybe that is the future model, because it's clear that the old model of a corporate R&D division just ain't working.When the founders of AT&T formed Western Electric and Bell Labs, they had no competition and little choice. They had to build their own equipment because there was no one else around, and they had to mount their own R&D because no one would do it for them. When Bill McGowan started MCI, he purposely did not start an R&D department because he believed that the industry had grown up - every vendor now wanted his business, and he could buy from the best.Today, both in the enterprise and carrier arenas, the incumbents (Alcatel, IBM, Lucent and others) have continually shot themselves in the foot, then attempted to solve the problem of having the wrong products by throwing a Hail Mary pass through acquisition - using financial engineering as a substitute for real engineering.One major problem when you build a company with cobbled-together technical platforms is that they just do not "hum" together. Integrating all these platforms results in continual delays, unanticipated costs and a need for even more staff, as companies are forced to become their own systems integrators. If the communication industry's laboratories had worked as they were supposed to, we wouldn't have this problem, but I suppose it was inevitable.