Visualizing the future by playing with data

Sharing public versions of tools I built and use in my research

In thinking about how I might model out what our future looks like in the year 2120, I like to have visuals that can show me a breakdown of key variables at once over a long sweep of time. It’s very helpful for visualizing the full picture of what the possible trajectories might be.

Seeing what the magnitude of change can be over 150 years makes predicting 100 years at least somewhat less daunting of a task.

Sometimes, great tools exist already. This is particularly true for economic and financial data— The Federal Reserve Bank of St. Louis has a plethora of datasets in FRED which can very easily be combined and manipulated to make charts like these:

A chart comparing what was more in a given year— wage growth (positive) or productivity growth (negative)

In other cases, datasets are amazingly comprehensive, but the visuals are lacking. This is particularly true of the occupational data from the Census that I have been working with over the past year, which I obtained from some moderately straightforward queries here (IPUMS login required).

Employment in the skilled construction trades has been remarkably consistent over time at around 4% since 1870.

I hope to share the tool I use to generate these charts about jobs more publicly one day, but I was able to improve my Google Data Studio dashboard I shared a few weeks ago:

More highly automatable jobs were more likely to see higher losses in the last recession (2005-2010), but this isn’t always the case.

You can now make charts like these yourself here comparing employment growth, wage growth, hours growth, the probability of automation (as rated by willrobotstakemyjob.com), and the percentage change in the gender ratio of each job.

Make sure you select at least one filter though— I recommend starting with time period.

Oh, and one more thing…

Sometimes, data is nigh inacessible to a non-technical user, or the summaries that are available are overly general. This has certainly been the case for data from the American Time Use Survey. But I am very excited to share a new Google Sheets tool I built which you can use to see what the average day looked like for a particular demographic slice over time.

Thirtysomethings are spending more time on work, sleep, and childcare, and less time on eating, chores, and socializing than their ancestors in 1965.

You can even see how much time your chosen demographic spends on a particular activity compared to other people, and how each generational cohort has differed with the passing of time!

People generally spend less time socializing when they get older, but nearly every generation has been less social than the previous one.

I have really enjoyed playing with this tool. Economic data can sometimes be abstract and the possibilities over the course of 100 years can seem endless. But we will still have 24 hours in a day (at least on Earth!) in the year 2120, and it’s grounding to consider how we might ideally spend our time.