Koby’s Kaos

New Neutrino experiment at Cern repeats the impossible

According to an article by Jason Palmer, a Science and technology reporter with the BBC News (http://www.bbc.co.uk/news/science-environment-15791236), the latest rerun of the Cern experiment again found that Neutrinos traveled through 700km of rock faster than possible before reaching Gran Sasso’s underground laboratories.

The new work used shorter bunches of neutrinos, to comply with earlier criticisms.

Their latest experimental results have been posted to the Arxiv repository and submitted to the Journal of High Energy Physics, and have not yet been reviewed by the scientific community.

The experiments were carried out by the Opera collaboration (Oscillation Project with Emulsion (T)racking Apparatus). The initial experiments, 15,000 separate measurements spread out over three years, found that the neutrinos arrived 60 billionths of a second faster than light would have, travelling unimpeded over the same distance. The latest results echoed those results.

Next year, teams working on two other experiments at Gran Sasso–Borexino and Icarus–will begin independent cross-checks of Opera’s results. The US Minos experiment and Japan’s T2K experiment will also test the observations. It will be several months before they can report.

I have been assured by a top theoretical physicist that these results are impossible and that something is wrong with the experiment. I then consulted with myself in both the future and the past, and we agree. This just can’t be happening and that practical time travel will never occur—except when filling out time cards.

Sometimes it’s amazing that anything works. Take the recent “Occupy” demonstrations (and let’s not extend it to the “Arab Spring” unless we have to). Any given metropolitan police department will have good, if complex, interior communications and a proven chain of command. Any given small group of demonstrators won’t, but can stay coordinated, kind of, within yelling distance.

However, when you add social media to these diffuse demonstrations, they suddenly become scalable to paralyzing size, though headless or million-headed depending on your point of view. To scale up near as far, police need complex additions to their communication, legal protocols, cross-jurisdictional protocols, command protocols, interagency training, and lots of time. Swarms happen and will be happening far more often, while policing has to be planned or else the reaction to unanticipated scaling becomes disabling inaction (Watts) or violent reaction (Chicago 1968).

Following the analogy, chip Interface IP can (usually will, but not should) be specific—efficiently planned to handle anticipated demands, or general—relying on design standards and conventions that will be far less efficient, but can be more elastically applied. Think of it as String Theory for chips, the universes get smaller but their clouds of possibility continue to grow.

Of course, a complex IP that leave no room for the actual functional IP would be a non-starter. Social Media for IP would have to be either a Fuzzy “applier,” EDA that sizes down to fit the specifics and keeps updating as the SoC design develops, or a Fuzzy chip interface that adapts to a probable/possible range of needs. The final SoC would be frozen in time unless it was something like an FPGA. In which case my Fuzzy Programmable Gate Array (FZPGA for short) would continue to have adaptability to handle a staggering matrix of unintended special cases as required.

Or not. Admittedly I’m sketching away on a napkin and trying to think of a better way to get you to pay for the coffee and scone. Remind me to tell you about my plan for 4^th Dimensional stacking in a chip or device. It’s verging now on possibility and is, I think, inevitable.

Maybe. But how many billion are you willing to bet?

Once upon a time I worked with a company renowned for engineering excellence, and for building complex new things just because they could. The idea was that once industry knew it could, it would want to, bad enough to pay for the fastest and best. In the technology heady days of the Cold War, that wasn’t such a bad idea—every battle axe needed a cutting edge. Brute force was an acceptable design principle.

Now, in a Jobsian economy, things are different. We are blessed with markets that are willing to pay a premium for an edge in style, utility, and speed. We are also cursed with a spongy economy featuring vanishing funds for schools and real research (and the specially-developed tools required).

Consumers, bless their chip-enabled credit cards, are now the market drivers, not the search for the sharpest sword or the hardest shield. Reminds me of the hey-day of Detroit, with bewildering new slabs of sheet metal rolling out every September—whether there was anything worthwhile inside or not. (A frightening analogy for today’s trendy tech, but that is another rant.)

Contemporary style has its own altars: lighter, smaller, faster, easier. Yesterday’s technology, even if it is practically given away in thrift stores, is only good for lining virtual bird cages. To even have a chance for inclusion in tomorrow’s winning mass market products, new IC’s demand tinier processes with bludgeoning development costs.

New fabs cost billions, yet they must be built on the hope that they will match future needs and be fully used once they come on line. Fabless designers swirl cells together and hope the new fabs can produce their dream circuits. And they all hope the total package will compel consumers to keep buying.

So what can you do?

I think you have to design for the future. As much as you hope for a home run, you build your game on bunts and singles. You improve your processes and tweak your costs. Your turn your clever ideas into reusable IP and build software architectures that dictate good practices for maintenance and expansion. You get as much out of existing technology as you can. You design for manufacturability and yield. All those traditional things one does to hold costs down and stretch the benefits of your research. You assume your competitors are doing the same thing and try to improve continuously, no matter how much Dilbert scoffs.

You need to keep the business going while you assemble your war chest. Why? Because you will have to jump off the cliff soon enough. Not many fans buy tickets to watch defense. They come for the home runs, and the crashes.

To increase your chances of one over the other (why does Wiley Coyote come to mind?), you have to invest in research strategies as much as the research itself so that you will have a profitable place to land. I am not talking about hiring away talent or stealing secrets (which is done with cold war dedication in our very scary world). I am proposing more research consortiums that unite schools and companies to share the resources and risks of research. These partial partnerships can keep costs down and imaginations vivid. They also can keep both the companies and the schools current with industry bests. When the companies feel the research has reached their strengths they can bug off and privatize their special sauce, but the research keeps bubbling, brewing better engineers and scientists as well as birthing new gourmet companies with an innovative product and a reasonable chance of success. Combined research can create a better technology ecosystem, one that will continue to ride and respond to changing demands and possibilities.

(And keep technology editors gainfully employed explaining them. Classic Win-Win!)