Innovation is no longer about building a better mousetrap. It’s about building better-armed cats.
That was the first thought I had when I read late last week about Linaro[1], a new company formed in collaboration by ARM Holdings, Ltd. and five of its licensed semiconductor manufacturers (”semis”): Freescale, IBM, Samsung, ST-Ericsson and Texas Instruments.
Linaro promises to provide a “common software foundation” for ARM-based Linux distributions. Their stated goal is:
To make open source development easier by delivering a common software foundation across multiple distributions and vertical segments, deployed by the industries leading OEM, MNO, and leading semiconductor companies.
On the surface, as some bloggers have pointed out, Linaro looks like yet another loose industry conglomeration that promises to end fragmentation and promote a new era of goodwill among competitors. We have certainly seen this in the mobile world over and over again, with LiPS, LIMO, etc. However, there are a few differences with Linaro that make it worth watching.
First, Linaro is sponsored by a group of hardware companies, specifically ARM Holdings and five of its major licensees. Not software manufacturers. That makes a difference because the output Linaro has promised is software, not hardware, and free, open-source software at that. Semis have always offered reference software with their processors, but they have not historically created new companies to improve the quality of that software, so this is a new entity, and its role in the marketplace will evolve over time. The primary focus, though, is on software interfaces to hardware, which will definitely benefit developers worldwide.
Second, Linaro is not building a better mousetrap - they are not creating a separate ARM distribution. What they are doing is forming an alliance of all the big cats in the neighborhood and addressing their combined weaknesses. The mouse in question (Intel, as Om Malik points out) has a great ecosystem in place, and the consignees for Linaro have just created a unique alliance for the cats.
Third, Linaro is not a traditional consortium like GENIVI, nor a de facto oligarchy like Android’s OHA (a benevolent dictator is still a dictator). Nor is it a standards organization like LiMo. Instead, Linaro was formed as a distinct company, not a consortium, from the resources of its constituent members, which gives it a strength and individuality that a consortium can’t match. GENIVI, OHA, and LiMo would not exist without their parent organizations, but Linaro could conceivably break off from its parents and strike out on its own someday. (I did note that nothing has been announced with regard to monetization, which would have to come first.)
Our ever-insightful friend Bill Weinberg has pointed out this morning that Linaro aims to be “the glue between silicon and software”. That metaphor captures the essence of Linaro’s goal in terms of filling gaps and sticking together important pieces of a puzzle, although Linaro isn’t actually providing any new hardware glue - all of those companies are already good at producing drivers, kernel patches, etc. Linaro’s goal is rather to fill software gaps among various distros that come directly from semis, enabling them to spend fewer resources on the kernel patches and drivers, and differentiate themselves lower down the stack in hardware. That isn’t so much gluing software to silicon - it is combining resources to glue business competitors into a unified front against the big mouse on the block.
The four things Linaro promises on their wiki are tools, kernel, middleware, and validation suites. Development tools sound awesome, and they are a great place to add value. Kernel patches and drivers are always good, although reasonably motivated developers usually find a way. Middleware integration is key, of course. But I think the kicker is in validation - stability will rule the day, as it has done with Android. Moreover, this also provides a lot of opportunity for companies like MontaVista to add further value, as we do with Android testing in our recently-announced Android Rapid Development Program.
There are four questions that remain vital in my mind:
As a Developer’s Advocate, I approve. And I’m anxious to see how the cats get along.
[1] For those wondering, Linaro is a horse - not just any horse, but a prize Holsteiner stallion. Born in 1989, he is the genetic ancestor of a number of “approved” (read: award winning) stallions.
