News

From the Dean: About technology development

The progress of technology is closely coupled to scientific discovery and invention.

The role of serendipity is often underlined, and the crucial value of basic, curiosity-driven research is emphasized. One often quoted example is the laser. While its ideas rose from basic research in quantum optics, its widespread applications range from medical technologies to telecommunications and mechanical engineering.

Even if chance favors the prepared mind, discovery and invention appear to be unpredictable processes. Whether a given discovery leads to technological progress, requires effort and investment and a fair amount of luck. Therefore it makes sense to focus on creating and accumulating basic knowledge, with an open mind for innovation and applications.

Thus it is somewhat surprising that a general pattern in the historical data of technological progress has recently been noticed. The pattern is well known as Moore’s Law in microelectronics. For several decades, the density of transistors on integrated circuits has doubled every two years, and the computational speed has advanced even faster.

This exponential pattern holds in fact for many technologies, from batteries and LEDs to cars and nuclear power. The exponential improvement holds for these and many other technologies, but the timescales for doubling can be quite different. The annual improvement rates vary from 3 to 65 per cent, according to a recent study by C.L. Magee of MIT.

The interesting corollary of this is that it may be possible to make predictions of the future technological progress using historical data, instead of relying on the forecasts of experts in any given field. Doyne Farmer and Francois Lafond have developed a model for technology improvement based on the empirical observation described above. Their model for technology development is a stochastic process, a random walk, which naturally shows exponential long-term growth. Moreover, the mathematical properties of the random walk make it possible to obtain estimates for the distribution of forecast errors. For example, their method suggests that the price of photovoltaic modules will most likely continue to drop 10% annually, but that there is 5% chance that the prices in 2030 will actually be higher than today.

One really does not know why Moore’s Law holds for many individual technologies. The fact is just an observation based on empirical data.  The interesting question is what are the mechanisms than make one technology grow faster than others.  There are suggestions that technological advance is similar to an evolutionary process. Biologists know that some organisms evolve and adapt more rapidly than others, due to their ability to fast experimentation by altering some elements without destroying the underlying functionality.  Among technologies, the faster evolving ones seem to be simpler in the sense that they have fewer interdependencies between their elementary components. The surprising conclusion then is that complexity in technology slows discovery and advance. It also follows that older technologies do not necessarily grow more complex over time.

Risto Nieminen
Dean, Aalto SCI

  • Published:
  • Updated:
Share
URL copied!

Related news

KTT Sami Itani
Research & Art Published:

Doctor’s career path: CEO Sami Itani’s story

‘Doctoral studies consist of team-based project work in which you can act as your own project manager.’
kaksi naista istuu punaisilla tuoleilla studiossa keskustelemassa, taustalla musta verho ja keltainen lattia
Research & Art, Studies Published:

What can we learn from the pandemic?

Aalto University's studio discussion focused on what the COVID-19 pandemic has taught us about the importance of the environment for health and wellbeing.
Assistant Professor Peter Nyberg, McKinsey Teacher of the Year 2020.
Honoured, Research & Art Published:

Aalto BIZ Teacher of the Year 2020

Finance Assistant Professor Peter Nyberg has been nominated as the Aalto BIZ Teacher of the Year 2020.
Machine Learning Coffee Seminar logo in purple and white colours
Research & Art Published:

Machine Learning Coffee Seminars continue with a great lineup of speakers

MLCS gathers together people from different fields of science who have at least one thing in common, interest in machine learning