5 ms·
>Why were either allowed to become laws. Moore's law was a little confusing in high school Nobody allowed Moore's Law to become a law. The so-called Moore's L
by jasode 6y ago
>Why were either allowed to become laws. Moore's law was a little confusing in high school
Nobody allowed Moore's Law to become a law.
The so-called Moore's Law is really just Gordon Moore's Extrapolation of an observation. In 1965, he observed that transistor count was doubling every ~12-18 months. He extrapolated that this trend could continue for ~10 more years to 1975.
But journalists and readers like snappy poetic words. "Extrapolation" is a long-winded 5-syllable noun so it's no surprise that the 1-syllable "law" was used instead. Wikipedia says: shortly after 1975, Caltech professor Carver Mead popularized the term "Moore's law".[1]
Although calling it "Moore's Extrapolation" would be more accurate, it just doesn't roll off the tongue as nicely. Sometimes humans value the sing-song poetry of a meme more than its dictionary accuracy.
[1] https://en.wikipedia.org/wiki/Moore%27s_law#History_of_the_concept https://en.wikipedia.org/wiki/Moore%27s_law#History_of_the_c...
- dathinab 6y agoOr Moore's Assumption. Which is less of a mouthful, through also less precise.
- vezycash 6y agoMoore's observation.
- sushshshsh 6y agoMoore's induced conclusion!
- ghaff 6y agoAbsolutely right. An observation that became something of a self-fulfilling prophecy.
- ordu 6y agoI'm not sure about "self-fulfilling" part. If Moore didn't do his observation, would doubling of number of transistor every 18 months still work?
- ghaff 6y agoI don't know for sure but Moore's Law was part of a certain expectation that was set for the industry and the market. For better or worse, this expectation tended to force various decisions that kept advances close to the predicted line.
- ncmncm 6y agoThe expectation drove capital investment. The business application of it meant profits from each generation could fund the higher costs of each succeeding generation. Now we are in the tens of $billions for a new generation, with much smaller increases in performance, and it is faltering. Will we spend hundreds of $billions on factories for smaller transistors? Or will we find a more cost-effective way to get more performance?
- ghaff 6y agoIt appears as if chipmakers will continue to drive things as small as they can, e.g. with EUV (finally). But the expectation is that a lot of continued performance gains will have to come through other means. These are some of the things that were talked about at the Hot Chips conference: https://enterprisersproject.com/article/2020/9/moores-law-what-means-today https://enterprisersproject.com/article/2020/9/moores-law-wh...
- ncmncm 6y agoCarver Mead often gets credit for Lynn Conway's work. If it's about device physics or transistor and single-gate layout, it is likely Mead. If it's about organizing logic functions on a chip, it is more likely Conway. She relied on Mead for a lot of the publicizing of her design methodology, because he was better on stage.