Interview: IBM's Linux chief tech officer Paul McKenney

IBM's chief technical officer of Linux Paul McKenney. Photo / Supplied

On February the 19th, Wellington will be awash with some of the greatest and brightest minds in computing as the Multicore 2013 Conference kicks off at Wellington Town Hall.

I caught up with one of the keynote speakers, Paul McKenney, who is IBM's Chief Technology Officer of Linux and got to pick his brains on the rise of Linux and machine intelligence.

PP: What originally got you into Linux?
PM: My first involvement with Linux was in 1997. I received an email from some guy I had never heard of with a .ru email address, who was asking for a machine-readable copy of one of my old papers. I sent it to him, and asked what he intended to do with it. He replied that he would add it to this kernel named "Linux". Although I had heard of Linux, it would be some years before I understood who Alexey Kuznetsov was.

PP: What do you see as the key compelling factors of Linux?
PM: Linux serves a great many uses and is not tied to any specific hardware. Because it is open source, anyone who needs it to do more or run on different hardware is free to do so. This freedom is extremely important, in fact, this freedom is exactly what enabled Linux to grow to serve so many uses on so many different types of hardware. It is not simply that many servers run Linux, and it is not just that Android smartphones (and a few others!) run Linux. It is that many common household appliances run Linux.

PP: How about your views on the future of Linux?
PM: Linux's base capabilities will continue to expand: More scalability, better energy efficiency, improved real-time response, and more and better applications support. Application support will be especially strong for the Android flavours of Linux.
In fact, there might come a time when many servers need to support the Android environment, and I would expect that smartphones and embedded devices will continue to drive important features and functions into Linux.

PP: There's a lot of different OS's out there all vying for attention, which seems to be creating quite a bit of consumer confusion, does the multitude of Linux distros ad to this?
PM: Actually, I suspect that very few consumers know, care, or need to care what form of Linux their smartphones and household appliances are running. Of course, for laptops and desktops, some care is still required. That said, Steam recently ported their environment to Linux, which seems quite likely to lead to an explosion of mainstream games becoming available on Linux. If this happens, it would remove a key barrier to Linux adoption.

PP: Which distro would you recommend to the average user? What about power users?
PM: If you are an average user considering running Linux on your desktop or laptop, your best bet is to run the distro that your friends and acquaintances are running. That is what I did at a particular point in time, so I run Ubuntu, which is not a bad choice. However, had I switched at some other time, I might well be using some other distro. If you are a power user, you probably have already made your choice, and will stick with it no matter what I say. ;-)

PP: Machines such as Watson seem to be getting increasingly smarter, do you think it is possible for machines to achieve human-like intelligence?
PM: I believe that machines will continue to get smarter. But I also believe that we humans will be able to avoid obsolescence for quite some time. For example, Watson is being adapted to medicine, but is designed to advise human doctors, not to replace them.

PP: What technology challenges do you see to building machines with human like intelligence?
PM: Energy efficiency. Yes, Watson did prevail in Jeopardy, but it consumed quite a bit more power than did its human opponents. Common sense. Although Watson seems much more human than Deep Blue (IBM's chess-playing computer from the 1990s), Watson nevertheless is very specialised. For example, Watson would not be able to safely cross a busy street. Yes, we do have self-driving cars, but impressive as they are, they would have absolutely no clue what to do if they were on Jeopardy. "Theory of mind", or guessing the user's intent. Although we are seeing some important advances here, for example, smartphones that guess what words you are going to write next, current capabilities do nothing more than scratch the surface.

PP: Could machines become self-coding, in effect learning?
PM: To some extent, they already are. A number of languages have "just-in-time compilers" (JITs) that reshape the software to better accommodate changes in the machine's environment and workload.

PP: I heard Watson got hold of the Urban Dictionary and developed a rather "ripe" vocab, it must be hard to teach an AI language context..
PM: Anyone who has raised children will have noticed that many children have a natural affinity for "ripe" vocabulary. Many are the parents who have discovered to their embarrassment that if you say a word with enough emphasis, you only have to say it once in order for your child to learn it all too well! So the question for you is whether you believe that this is a sign of Watson's improving levels of humanity. ;-)

PP: Aside from TV game shows, what sort of applications could you see for such machines?
PM: Medicine is a huge area. Currently, Watson is learning about oncology, starting with lung and breast cancer. Although I firmly believe that Watson's contributions will be extremely valuable and that Watson will help save many lives, there is far more medicine that it does not know than medicine that it does know.

PP: So IBM's Linux CTO, what do you see in your crystal ball for IBM going forwards?
PM: As IBM's CTO for Linux, I see IBM working with a greater variety of open-source software. Interestingly enough, Watson is an example of this, as it is built on top of a number of open-source projects. That said, I expect that the Linux kernel will remain my first love for quite some time to come.

PP: What about the wider IT industry?
PM: I do remember the supercomputers of my youth quite fondly. Nevertheless, there is a huge amount more computing power in the typical smartphone than the celebrated Cray-1 could muster. Internet bandwidth has similarly increased. This computing power and bandwidth will open up new vistas of computing and even of life itself that we cannot hope to even begin to imagine. This is therefore a wonderful and exciting time to be in the computing field!!!