There is innovation bottleneck that is partly cultural and partly human resource constraint.
Advanced statistical techniques, data science and simulation resources available have increased the potential what is theoretically possible to discover and reason with, but it's hard to use them in full capacity.
We live in era where something like probabilistic programming has dramatically increased the complexity of models that can be used for reasoning, science is still struggling with the meaning of p-values and publishing the open data used for others. In many areas researchers cant use techniques that peer reviews are not very familiar with.
Scientific discoveries require increasing amount of resources and collaboration and methodological knowledge. Some biochemists spend 5 years developing computational package for Matlab and have to learn completely new fields atop of their PhD. The amount of combined knowledge and methodology to get totally new discovery and not just incremental advances stretches what small research groups can do on their own. You have to build whole frameworks from scratch.
Big money places like Cern have money and people to pull it off. They can employ 100 physicists to work 5 year for software package that analyzes some experiment and squeezes the science out of data.
TL:DR: method knowledge available >> method knowledge used.
Advanced statistical techniques, data science and simulation resources available have increased the potential what is theoretically possible to discover and reason with, but it's hard to use them in full capacity.
We live in era where something like probabilistic programming has dramatically increased the complexity of models that can be used for reasoning, science is still struggling with the meaning of p-values and publishing the open data used for others. In many areas researchers cant use techniques that peer reviews are not very familiar with.
Scientific discoveries require increasing amount of resources and collaboration and methodological knowledge. Some biochemists spend 5 years developing computational package for Matlab and have to learn completely new fields atop of their PhD. The amount of combined knowledge and methodology to get totally new discovery and not just incremental advances stretches what small research groups can do on their own. You have to build whole frameworks from scratch.
Big money places like Cern have money and people to pull it off. They can employ 100 physicists to work 5 year for software package that analyzes some experiment and squeezes the science out of data.
TL:DR: method knowledge available >> method knowledge used.