A Three-Day Week
What happened to the 24-hour working week we were promised in the 1960s?
Public figures at the time were deeply worried about “the great intimidating fact of expanding leisure”. Working hours had fallen from 70 a week in the 1900s to just 40. The problem was even debated in the House of Lords and it makes for interesting reading. War hero Field Marshall Montgomery suggested that schools should “teach the young people of to-day how to use their leisure time as they grow up”. People in the 21st century would be “truly perplexed” by the vast stretches of free time they found themselves with.
I don’t know about you, but I’m not feeling perplexed. I’m feeling busy. We’re still working 40-hour weeks, we’re commuting further, and we have succeeded in finding more ways to busy ourselves during our leisure time than anybody imagined in the 1960s.
Robots
So why don’t robots do everthing for us? Well, they do. Manufacturing productivity has caused a fundamental change in human life in the past 250 years. Taking Adam Smith’s classic pin factory example, a worker who could produce 5,000 pins a day in 1760 can now make 800,000 a day, and spend fewer hours doing it. Factories were once workshops populated by toiling humans; now they’re labs full of gleaming robots.
People think that developed countries are no longer manufacturing economies, but that’s only partly true. The contribution of manufacturing to GDP did fall in the UK between 1950 and 1990, from 37% to 21% (it’s 19% in 2019). But that’s not because we’re not making as much — it’s because we’re making it so efficiently that it’s much cheaper. If you adjust for price changes, manufacturing’s share is only three percentage points lower (24% down from 27%). We produce almost the same amount of stuff, but the same items cost less than they used to. And since time is money, that means we’re spending less time creating things than ever before.
Services
So what are we doing with our time? The Office for National Statistics can tell us, at least for the UK. They’ve split who does what out into sectors.
It turns out that much of our time goes into keeping each other healthy, informed, and safe — health, education and defense is the highest chunk of employment, along with other Government functions. The next biggest area of work is professional and support, which covers a wide range of activities from advertising and management consultancy to legal services — so very broadly speaking, ‘organising everything else that goes on’. We also spend more of our effort on distributing goods than we do on actually making them — retail and wholesale is 13% of employment compared with 8% for manufactured goods.
Knowing this, it’s no surprise that service economies like the UK and the US are said to be in a ‘productivity crisis’. The jobs their workers do can’t easily be automated. It’s easy to see how you can scale up pin production, but how do you enable a lawyer, doctor or software engineer to do their job 40 times faster?
Software
The answer lies in the 4% bar near the bottom, the ‘info & commmunication’, part of which is software engineering. Software engineering is different from other service activities, it’s how we automate service activities.
If you’ve ever received customer service in an online chat box rather than over the phone, you’ve already benefited from this. With online chat, customer service staff can handle multiple queries asynchronously. Often a chat bot will handle at least some of the call as well.
In law, the use of AI to automate thinking is beginning to take off. Services like Lawbot automatically examine free-text contracts and give advice on how to improve them — I pasted in a standard software EULA and got tips such as “Ensure that the definition of Software includes updates, releases, bug fixes, and enhancements thereto.” The Do Not Pay app claims to be the “World’s first Robot Lawyer.” You give it the basic facts of your case, it constructs the necessary paperwork, and even generates a script for you to read from in court. This is pretty advanced stuff, and it’s changing the world by slashing the cost of services.
But it doesn’t always take AI to automate brain work. Sometimes the answer is simpler than that. Much service-industry work can be split into ‘creative activities’ and ‘mind-numbingly mechanical mental drudgery’. Even studying for a PhD can be 5% pushing the boundaries of human thought, 95% pipetting fluids into petri dishes. There’s still a huge need for straightforward automation.
Tooling
Software engineers are realising this more and more. A revolution is taking place in software tooling that reveals how difficult we’ve been making things for ourselves up until now. Ten years ago, getting anything running on a new machine was an effort that required a human brain and some deal of time. Now you can just spin up a Docker container and get going. Ten years ago, companies employed entire teams to manage basic build and deployment infrastructure. Now you can give CircleCI a config file, point it at your GitHub repo and let software magic do the rest.
Ten years ago, building a web app required wrestling with HTML, CSS, JavaScript, PHP, Apache config, iptables, and dozens of esoteric technologies with their own configuration languages. Platforms like Anvil unify all this under a single philosophy — Python everywhere, even in the browser.
These platforms take away the 95% of time you spend metaphorically pipetting fluids into petri dishes and allow you to spend all of your working day pushing the boundaries.
So invest in your tooling. Automate and simplify everything you can. Work at the highest abstraction level possible. Do whatever it takes to eliminate the drudgery and focus on the creativity. And who knows, maybe one day you’ll work a three-day week.
