The 2x workerPosted: April 29, 2012
Not everyone will be a super star, but being tech-able is one thing that has allowed me to considerably improve my performance. I started to code a few months ago, and I’m amazed at how much I’ve improved as a worker since then. In this post I’ll try to break down how:
10x vs. 2x
Superstars are those who have a small advantage over their rivals but are a lot more valuable. Imagine [ a sports game] one on one against [really famous player] and [not famous professional]. Chances are you wouldn’t stand a chance, either way. Yet typically, the famous player earns a lot more than the average pro, because he is a lot more valuable to the team.
When I started to get interested in programming, I noticed that in software development, a lot of posts highlight how you should focus on superstar economics: to get 10x performance hire the 10x developers, hire the 10x teams etc. That was a bit disheartening – why learn a skill that wasn’t useful unless I was amazing at it?
I’ve actually found that very quickly, and very easily, I’ve been able to apply some python code to some hard (but not critical) business problems I had. Without the ‘python’ tool, I would have had to push the problem to our own developers, spend a lot of time on it, but most often I would have just given up and moved on. A crude estimate is that all in all my work now add twice as much value to as I did before.
Let’s look at an early example:
Fuzzy matching at scale
Part of our efforts at Rangespan are around correctly matching products from our suppliers to those already in our database of products. A small part of this process is correctly identifying manufacturers: one of our suppliers might tell us about a computer built by “Hewlett-Packard”, but another might have the manufacturer down as “HP”.
One way to make this easier is to look at a name we don’t recognise, and look at the ‘closest’ match, which might help: ‘vaio’ looks a lot like sony-vaio, for instance. There is a way to do this with a small program and very little programming skill:1. open a file with a list of unmapped manufacturers 2. open a file with a list of all mapped manufacturers 3. for each unmapped manufacturer 4. find the closest mapped manufacturer 5. output a file with the unmapped manufacturer, mapped manufacturer, and the distance between them.
1,2,3, and 5 are all really easy to do when you’re only half way through the great Learn Python The Hard Way. Step 4 sounds hard, but it was easy enough to find an example of how to do it online.
The Real Benefit
The direct benefit is obvious – I can solve problems I couldn’t solve before, and I’m faster doing some of the same things I was doing before. This only affects roughly 20-30% of what I do though. There are several other things that come into play:
- I enjoy these menial tasks more and do them more readily.
- I have a better handle on the technology I use, and I feel more comfortable with customisations and improvements in anything technical.
- I can better understand the kinds of things technology can solve and the kinds of things that need human intervention.
- I can speak to developers and understand more of what they say in response. This means less friction when we work together, which is a huge bonus.
Since that first example, I’ve found coding to be a great help in several other areas – and I still haven’t fully finished learnpython yet. I’ll add to this blog as new kinds of things happen and update this post with feedback. I’d be particularly keen to hear from anyone who’s had the experience of succeeding in a business role by using programming techniques and insights.
—————Notes: Avichal has a good example of applying superstar economics to selecting your team. There are probably better ways to do the example above. Great – but that’s not my point. My point is that the method I found works well enough to be useful. I’ve been learning by reading Learn Python The Hard Way. When I say ‘distance’ in the example above, I mean this. I found this example of how to calculate that distance using python. The numbers I use to calculate efficiency are completely made up.