At 8-years old, Andrew "Bunnie" Huang appreciated the fact that his Apple II came with schematics and source code because it allowed him to figure out how it worked.
"I was wondering what all these little black things on the board were and I would take the chips out and put them in backwards, even though my dad told me not to," said Huang during his EE Live! 2014 keynote on open-source hardware and the future of embedded systems. "He was right; you don't put the chips in backwards."
Today that information is guarded and protected in the hardware industry and Huang, now a research affiliate at MIT who holds a PhD in electrical engineering from the school, realized this change wasn't because hardware became too complex, but because it was too easy to improve, and Moore's Law was tough to keep up with.
If Moore's Law saw technology doubled every 18 months, that meant someone working on a linear improvement, like optimizing a process node, could be getting 80% performance improvement per year, and Moore's Law would be shipping something better by year two.
"So the problem has been that sitting and waiting has actually been a viable strategy versus innovation," said Huang. "This problem is particularly acute in hardware."
Moore's Law catches linear improvement in year two. Huang's EE Live! presentation is available here.
In software, the innovation cycle to innovate, distribute, and adopt a new product can occur in a matter of days to weeks, but hardware takes longer.
"Distribution is really the killer," said Huang. "In software, when you want to push a patch you copy to the server and you're done. But if you want to get your [hardware] out there, it can take months to years -- an eternity in software."
That means the software innovation cycle is well within a generation of Moore's Law, but for hardware the cycle can take longer than a single generation of Moore's Law, so the system has favored really big businesses.
"You can't just design a product and be a successful business, you design a series of products all at the same time," said Huang. "You have a pipeline with two to three products in the pipe, or else someone is going to beat you to the punch. So you need multiple teams of people and a lot of money to fund them."
Huang believes this has led to a veil of secrecy in the industry, because even though the details of hardware can't be hidden, reverse engineering takes time and companies use that to their advantage.
"The fact is it does take a few months to do the reverse engineering. If you're not shipping schematics and source code and you delay your competitor by a few months, you've actually bought a significant competitive advantage in the Moore's Law domain," said Huang.
The light at the end of the tunnel for those trying to compete in the hardware industry is that eventually Moore's Law will slow down. Huang believes that soon you won't be able to buy a faster computer, a flash drive that stores more data, or a smaller or more powerful phone every year.
"It used to be that $1000 yesterday bought you twice the computer a year from now," said Huang. "It may be bad news for Intel and a few other guys, but I think this is great news."
Next Page: Slow development movement