“Moreover, I observe that machine-learning or model-based or data-analysis solutions on forecasting weather, pandemics, supply chain, sales, etc. are happily adopted, and the startups that produce them reach quite high valuations. When trying to explain why prediction markets are not adopted, this makes me favor explanations based on high overhead, low performance and low applicability over Robin Hanson-style explanations based on covert and self-serving status moves.”
I agree that the success of bespoke ml tools for forecasting negates some of the Hansonian explanations, but probably not most of them.
As ML tools replace human forecasts, they do not pose a threat to the credibility of executives. They do not have to provide their own forecasts that could later be falsified.
(Speculative) The forecasts produced by such tools are presumably not visible to every employee, while many previous instances of prediction markets had publicly visible aggregate predictions.
These tools forecast issues that managers are not traditionally expected to be able to forecast. Weather and pandemics are certainly not in the domain of executives, and I am unsure whether managers usually engage in supply chain and sales predictions.
These tools do not actually provide answers that could be embarrassing to executives, and for which prediction markets with aggregated human expertise could be useful. For example, machine learning cannot predict “conditional on proposal by CEO Smith, what will our sales be”. A good test for this explanation could be how many companies allow feedback to strategy proposals by employees and visible to all employees.
These tools forecast issues that managers are not traditionally expected to be able to forecast
The thing is, not really. Some of these ML companies offer predictions for employee retention or project timelines, which managers would in fact be expected to forecast.
“Moreover, I observe that machine-learning or model-based or data-analysis solutions on forecasting weather, pandemics, supply chain, sales, etc. are happily adopted, and the startups that produce them reach quite high valuations. When trying to explain why prediction markets are not adopted, this makes me favor explanations based on high overhead, low performance and low applicability over Robin Hanson-style explanations based on covert and self-serving status moves.”
I agree that the success of bespoke ml tools for forecasting negates some of the Hansonian explanations, but probably not most of them.
As ML tools replace human forecasts, they do not pose a threat to the credibility of executives. They do not have to provide their own forecasts that could later be falsified.
(Speculative) The forecasts produced by such tools are presumably not visible to every employee, while many previous instances of prediction markets had publicly visible aggregate predictions.
These tools forecast issues that managers are not traditionally expected to be able to forecast. Weather and pandemics are certainly not in the domain of executives, and I am unsure whether managers usually engage in supply chain and sales predictions.
These tools do not actually provide answers that could be embarrassing to executives, and for which prediction markets with aggregated human expertise could be useful. For example, machine learning cannot predict “conditional on proposal by CEO Smith, what will our sales be”. A good test for this explanation could be how many companies allow feedback to strategy proposals by employees and visible to all employees.
The thing is, not really. Some of these ML companies offer predictions for employee retention or project timelines, which managers would in fact be expected to forecast.