Thread by @dashunwang

special delivery for #IC2S2 - our latest in Nature Human Behavior: nature.com/articles/s4156…

Five years in the making. Accompanied by my essay in @HarvardBiz: hbr.org/2019/07/how-ne…

Findings may be relevant to a range of scholars, CSS, diffusion, #ScienceOfScience thread [1/n]

Diffusion processes impact broad aspects of human society. our knowledge is mainly limited to spreading in non-substitutive systems. Yet, many ideas, products and behaviors spread by substitution-to adopt a new one, agents often need to give up an existing one 2/

We find that, in substitutive systems, ranging from phones, cars, apps, and scientific fields, their early spreading patterns are characterized by power law scaling with non-integer exponents, in contrast to exponential growth customary in such processes. 3/

We used a large mobile-phone data to trace how new innovation replaces old ones, and developed a minimal model to incorporate new ingredients we discovered, which allows us to explain diverse growth trajectories. We hope these results/findings are relevant to different scholars 4

For physicists who study social systems, check out Supplementary Note 4 about the detailed model derivations - hope to contribute to pushing the predictive power of mechanistic models to fit to real data. 5/

For diffusion scholars, these results may inspire a reexamination of existing diffusion curves, either substitutive or not, to see if they follow exp or power law. Also hope these contribute to new empirical/theoretical evidence for substitutive dynamics. 6/

For #ScienceOfScience, there's been relatively few studies on the growth of a field, and we hope these results offer some initial empirics that are useful for future studies. 7/

For business audience/decision makers, I wrote in this @HarvardBiz piece to talk about how to contextualize these findings. Check it out!

hbr.org/2019/07/how-ne…

8/8

One click access to the paper: rdcu.be/bJgDO