🛹 jaron.club 🛹

all posts are collaborative g**gle docs

The Civic Oracle

collaborate and comment in doc

Futurology and CIVIC APIs

“Strictly extrapolative works (sic) generally arrive about where the Club of Rome arrives: somewhere between the gradual extinction of human liberty and total extinction of terrestrial life.

It is far too rationalist and simplistic to satisfy the imaginative mind, whether the writer’s or the reader’s. Variables are the spice of life.

Predictions are uttered by prophets (free of charge), by clairvoyants (who usually charge a fee, and are therefore more honored in their day then prophets), and by futurologists (salaried).”

Ursula Le Guin, The Left Hand Of Darkness

In what ways should CIVIC predict the future?

In the past, Hack Oregon has dabbled and ideated in many experimental spaces. One of those spaces involves making predictions about the future. As we go through this state change to the Civic Software Foundation, it’s an opportunity to re-examine some of these experiments.

I don’t think that the Civic Software Foundation should host specific, numerical predictions or projections about the future behavior of people in our APIs, though this is not a hard-line stance.

Why not?

CIVIC, as a project, and organizational mindset, does not see the possibility for people, or groups of people, as fixed. We see the possibility in both people and organizations, to collaborate and change in new and interesting ways, in a way that fundamentally changes their behavior.

Using data to make specific numeric predictions that fix people’s decisions and actions as constant isn’t in line with this belief.

Data has so much descriptive power for the past and the present. It can reshape our understanding of what has happened, and make us change our plans. The relationships that are shown and insight from that description can be applied to shape our future — but it doesn’t define it.

https://blog.lareviewofbooks.org/provocations/neophobic-conservative-ai-overlords-want-everything-stay/

In addition, as a former futurologist (salaried), I see the activity as fraught with risk. Actuaries developed quantitatively complex and precise models for the cost of retirement plans, but results varied wildly from the model predictions because they discounted our human capacity to extend our lifetimes (and misjudged interest rates vs. investment returns). Other models of data based prediction are similarly fraught — Nate Silver is a political oracle, until he’s just wrong.

There’s other reasons not to make predictions about the future, especially in our collaborative work with government. Mercer, my former employer, used to be the actuary for Oregon PERS and other governments, until they settled a lawsuit with the state of Alaska for $500 million for making mistakes in predicting the future costs of retiree medical plans, after which they decided the business was too risky.

I’m willing to entertain possible exceptions, but I think that the Civic Software Foundation should generally avoid being data-based futurologists.

Conclusions

Hard-line stance: No futurology in APIs hosted on CIVIC 

Middle-ground stance: Futurology in APIs hosted on CIVIC clearly marked as experimental, obvious distinction made in all data visualization and written content between fact and future forecasts

Possible exceptions:

  • Stated, credible plans of organizations
  • Short-term credible forecasts of processes that aren’t subject to sudden change

Examples

Elections

Not OK:

  • Forecasting future election results
  • Forecasting future campaign spending amounts

OK:

  • Calculating voter acquisition cost
  • Showing relationships between spending and election results
  • Making a model that is predictive for past elections and highlighting components have the most predictive power