Slashdot Mirror
Boffins Fear We Might Be Running Out of Ideas (theregister.co.uk)
Innovation, fetishized by Silicon Valley companies and celebrated by business boosters, no longer provides the economic jolt it once did. From a report: In order to maintain Moore's Law -- by which transistor density doubles every two years or so -- it now takes 18 times as many scientists as it did in the 1970s. That means each researcher's output today is 18 times less effective in terms of generating economic value than it was several decades ago. On an annual basis, research productivity is declining at a rate of about 6.8 percent per year in the semiconductor industry. In other words, we're running out of ideas. That's the conclusion of economic researchers from Stanford University and the Massachusetts Institute of Technology. In a paper published this week through the National Bureau of Economic Research, "Are Ideas Getting Harder to Find?", economics professors Nicholas Bloom, Charles Jones, and John Van Reenen, and PhD candidate Michael Webb, defy Betteridge's Law of Headlines by concluding that an idea drought has indeed taken hold. "Across a broad range of case studies ... we find that ideas -- and in particular the exponential growth they imply -- are getting harder and harder to find," the authors declare in their paper.
I wonder whether it's really a lack of ideas, and or worse engineering staff. I think engineers are, on average, less passionate than they used to be. For many people in the industry it's just a career now, and not a passion. Especially in large companies like Intel.
That means each researcher's output today is 18 times less effective in terms of generating economic value than it was several decades ago.
Assuming that the absolute number of transistors still matters, this math is ridiculous. A doubling of transistor count now means roughly 10 billion new transistors vs. a doubling in the 70s meaning maybe 10,000. So for 18x the headcount you get 1 MILLION times the transistors. A researcher is about 50,000 times more effective than he was in the 70s.
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
We're not running out of ideas. What has happened in CPU development is that we have made all of the relatively easy advances in transistor miniaturization, and further advances are becoming incremental as progress runs up against the asymptotic curves imposed by the laws of physics. Further advances in processing power are therefore coming to rely upon increasingly multicore designs and sophisticated caches, mainly because that's a less risky business proposition than investigating architectures other than the von Neumann and (occasionally) Harvard architectures.
It's also worth noting that most of the several orders of magnitude increase in processing power over the last three decades has been consumed by increasingly inefficient software as a way of keeping software development costs down.
Nature only provides so many free rides, and humans have proven themselves very good at exhausting them quickly. Ideas, even good ones, are always cheap and plentiful. It's a willingness to do hard (and therefore expensive) work that is in short supply.
Proud member of the Weirdo-American community.
Another thing we can outsource to AI.
Possibly not if Moore's law fails. However, this study arguably just shows the failure of focussing entirely on applied research. For each of their case studies they focus on one thing, such as Moore's law, crop yields etc. and conclude that it gets harder each year to drive the increases. This is because each of these areas is sticking to one fundamental approach and refining and improving it more and more which is clearly going to get harder over time.
What keeps the ideas coming is fundamental research which opens up entirely new approaches to solving problems. As the quote says "No amount of continuous improvement of the candle would have lead to the electric light bulb". Indeed the entire IT revolution owes its existence to the discovery of quantum mechanics and its application to understanding condensed matter physics. Without this applied researchers would be still be working on improving the valve.
The problem is that governments love to focus less on helping companies develop better widgets. The economic returns are almost immediate - or at least immediately obvious - and so useful to a politician seeking re-election. What they need to do is to put more money into fundamental research so that as fields run out of ideas there are completely new areas full of potential ideas to improve our lives in ways we cannot yet imagine. The problem is that the return on this investment is both uncertain and likely 50+ years away and the average politician has trouble caring about anything further away than the next election.
To bring us this information...
This gave me a good chuckle. It's always amusing when people from the left or right try and explain the other side and get it completely wrong.
Newsflash to every extremist on either side: no political ideology is completely wrong or completely right. They all get somethings right and somethings wrong, and the same ideology doesn't always work in every situation and every society.
Get over yourselves. As with most things, the best solution is often somewhere in the between what the extremists from either side espouse. Stop demonizing or regaling people based on their political preferences.
"That's the way to do it" - Punch
"Everything that can be invented has been invented."
~ Charles H. Duell, Commissioner of US patent office (1899)
Not true. Commissioner Duell never said that, and what he actually said was pretty much the exact opposite:
In my opinion, all previous advances in the various lines of invention will appear totally insignificant when compared with those which the present century will witness. I almost wish that I might live my life over again to see the wonders which are at the threshold. -- Charles H. Duell 1902