I had a great conversation with @kmac as she prepared this column, much of which could not make it into her 1000 words. So here are some more thoughts: https://twitter.com/bobkopp/status/1360224071250755584

The most critical ‘technology' in climate services isn’t the climate models — it’s the people who know how to interpret them and assess their decision-relevant uncertainties. The climate translators, in her paper’s words.

I am a bit more skeptical of some of the other technologies highlighted in her paper.

In particular, while weather-resolving climate models are interesting, — my expectation is that in many cases, they will turn some ‘unknown unknowns’ into ‘known unknowns’, but probably won’t lead to a substantial reduction in overall uncertainty.

Rather, climate risk management is going to remain a problem of iterative decision making under uncertainty; and even if we had perfect physical models, economics and emissions would remain first-order uncertainties in future climate projections.

(Your ability to predict future climate risk is never going to better than your ability to predict future cumulative emissions and future economic structure.)

There’s no out-of-the-box app for climate risk management — since climate change touches every decision with decadal and longer implications, climate considerations need to be integrated on an ongoing, iterative basis into long-term private and public planning processes.

The information needed to do so effectively will depend upon the details of how each organization works. Thus, again, the need for translators.

Private-sector climate consultants can be a useful part of the climate services ecosystem, provided they are using methodologies that are clearly documented and traceable to open science.

Firms boasting about proprietary methodologies should be viewed skeptically — black boxes are not science.

And there are fundamental scalability limits that should probably discourage VC investment in climate services. If there’s ever a big climate service firm, it’s going to look more like McKinsey than Facebook.

Moreover, for reasons of equity and justice, the climate services ecosystem is always going to need a substantial public element — just like weather data and forecasts, climate information should not be restricted to those communities able to pay for it.

At the same time, climate risk information has many more dimensions than weather information, and the relevant aspects will differ greatly by region, industry, and sector.

Thus the centrally led model of national weather services is probably not the right model, either. Public climate services will probably work best with a decentralized, networked structure.

That’s why I like the land-grant model @kmac mentions. And I’ll have a paper with much more to say on that topic out soon.