BI: You wrote: “Mobile devices aren’t consumption devices, they’re prosthetic brains.” The Common thought is that they are consumption devices, so, what do you mean by that?
Alistair Croll: If you need to check your phone to find out about a meeting, you’ve failed. If your phone understands that you need to know about a meeting, and tells you about it, then you’ve won. In the former case, the burden was on you (to check) and the phone was simply a reference.
I think they’re both.
The theoretical foundations of parallel programs have historically been defined using unstructured parallel programming constructs such as threads and locks. While these constructs suffice to characterize fundamental properties of parallel program execution such as determinacy, data races, memory consistency, deadlock, and livelock, we argue that they are inappropriate foundations for programmability and performance of parallel software. One can see an analogy with the Turing Machine abstraction which suffices to establish theoretical properties of computable functions, but is an inappropriate foundation for sequential software.
The Habanero Multicore Software Research project at Rice University (http://habanero.rice.edu) has focused on identifying orthogonal sets of structured parallelism primitives that can provide a foundation for improved programmability and performance of parallel software. In this talk, we summarize key primitives for task creation, termination, synchronization and isolation, and show how they can provide rich semantic guarantees for different classes of programs while also being amenable to efficient and scalable implementations. The benefits of structured parallelism will be motivated by illustrating how these primitives can help improve the effectiveness of data race detectors, compilers, and runtime systems, relative to the use of unstructured parallelism.
This is what Apple “invented” the idea of sliding a latch to open something. But because they were doing it on a computer they got to patent it. Probably it cost some effort to work out the code to create the image and so forth – although if it cost them millions their programmers are incompetent – even tens of thousands seems high for that particular coding job. But here is the point: Nobody gets to copy their code with or without patents. The thing they actually paid for is protected.
And perhaps it’s just a matter of time before open source takes center stage in mobile. Today mobile developers are just trying to get work done in a highly fragmented market. As the industry settles on norms for the kinds of services and infrastructure one needs to build successful mobile experiences, I suspect we’ll see open source stake its claim to the market, similar to what happened in the server market 20 years ago.
Analysis of the costs and benefits of IP protection shows there is a tendency toward overprotection of IP in all our societies, particularly in the areas of copyright and patents.
The analysis also shows the optimal rate of protection differs between countries and that it can differ across time as countries move through different stages of economic development.
The problems of overprotection are particularly acute for technology importing countries, including developing countries.
The analysis shows that for these countries, IP rights that are too strong will detract from innovation rather than promote it.
It’s not a question of protection, it’s how long and how broad that protection should be.
Hamish MacEwan 