As I See It, March 2011

By Nelson Morgan, Director

"Well, in our country," said Alice, still panting a little, "you'd generally get to somewhere else — if you ran very fast for a long time, as we've been doing."

"A slow sort of country!" said the Queen. "Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast."

– C. Dodgson (a.k.a. Lewis Carroll)

Sometimes it seems as if the Red Queen had it right in "Through the Looking Glass"; for instance, technology moves so quickly that just keeping up takes a great effort (unless you're 15, in which case it's effortless). And yet my experience has always been that it feels like progress is extremely slow when you are working within any particular technology. You may see what needs to be done very quickly (and perhaps code up the key elements in a busy weekend), but then come all the necessary details...

The impact of technology on society seems to operate in still a different time scale — slow as molasses and then suddenly, dramatically fast. Smartphones have been around since the 1990s, but then with the introduction of the iPhone, something reached a "tipping point," and sales and impact took off. I'm not qualified to say whether the recent upheavals in the Middle East, particularly in Egypt, owed much to social networking tools, but clearly they were extensively used. There was even apparently an intervention enabled by a technology I've been interested in for a few decades, automatic speech recognition. When the Egyptian government brought down the Internet, protesters were able to call a phone number that used Google tools (including speech recognition) to generate Twitter tweets, keeping the movement connected. Maybe I'm oblivious, but Twitter just seemed like entertainment to me before this event — now it seems like it can be an enabler of social change.

This "tipping point" property, popularized by Malcolm Gladwell, seems to apply to much of the progress that we see in computer science and related disciplines. Sometimes this is due to sociological phenomena, which can suddenly make particular technologies much more important; but sometimes it is really due to discoveries that drastically shift the research landscape. The latter may not be noticeable in the public sphere for some time, but in retrospect we understand how each basic advance has changed the nature of our inquiry. For instance, hidden Markov models (HMMs) and the fundamental associated mathematics were developed in the late 1960s at IDA in Princeton, and in the early to mid–1970s these methods were applied to speech recognition at CMU and IBM. It was not until the mid– to late 1980s that HMMs were almost universally adopted for speech recognition, but in retrospect the earlier work fundamentally changed research and development for this topic. And of course, here in the Bay Area we're aware of a more fundamental part of nature that has this property — tectonic plates pushing up against one another for months, years, or even decades, until finally there is a significant earthquake.

Once a game–changing event happens in research, how do we make progress? Much of what follows is necessary but (some would say) uninspiring incremental progress. As researchers gain understanding about the basic ideas, they begin to extend it and apply it. As computational resources expand, the opportunities for exploitation of the grand ideas expand as well. Eventually the big new idea becomes the standard approach, and as such needs to be questioned and might be ultimately supplanted by the next big new idea. It's particularly the role of the young to question any standard wisdom, and it's equally the responsibility of senior researchers to push back, since most candidates for the next big idea will fail; but some of them won't, and that's the great thing to look for.

How do we get to the next set of big ideas? Moving quickly is necessary but not sufficient; like Alice, you could end up in a place that looks an awful lot like where you started. I know of no foolproof recipe for research breakthroughs, but certainly success is often associated with an interest in really understanding what is going on and a willingness to forego short–term gains. Serendipity also plays an important role, to be sure, but investigation driven by curiosity is still one of the best ways to be prepared for unexpected discoveries. Although this concept tends to be unpopular with sponsors, there is plenty of evidence that investigator–driven and curiosity–driven research is where most "next big things" come from. Since we also have to satisfy the shorter–term goals of sponsors in order to keep our work going, finding common ground between the aims of sponsor and researcher is a critical challenge. This issue of the Gazette will provide a few examples of how ICSI's researchers, past and present, have met this challenge.