Product Entropy

Welcome another long read. This one explores disorder.

After watching Tenet a little while ago, I realised three things:

  • I know very little about thermodynamics
  • “We live in a twilight world, there are no friends at dusk”
  • Entropy is a thing and it’s interesting

What has this got to do with Product Management?

Well, it’s triggered a few thoughts.

Sadly there are no answers to the inversion of time in this post. Much as we could all probably use that.

But, as you’ll gather if you read this, I’ve become slightly fascinated with thermodynamics, much of which forms the plot of Tenet. Reading up on some basic physics has made me realise how little I know about how the world hangs together.

As Product Managers we focus so much on understanding humans, our needs, our preferences. We learn about businesses, revenue and objectives. Not so much on the universe, matter, force or energy. Unless you’re a PM working for NASA or JPL or some suitably science-y organisation.

Yet the laws of physics are universal, by definition, and they can be applied to product management.

Anyone that’s watched this sci-fi time travel espionage epic from Christopher Nolan knows that you might have to watch it a few times before it starts to make any sense at all. Much like my videos on product stuff…

However, if you have watched Tenet a few times you’ll know all about its main star, entropy.

In this post I’m going to go into why us product people need to know more about this thing called entropy.

Cast yourself as the ‘protagonist’. As a humble Product Manager, that shouldn’t be too hard for you modest lot.

A typical day in the life of a Product Manager

What on earth has time travel and entropy got to do with Product Management, you might be mumbling.

Well, roll with me, and I’ll tell you.

Let’s create entropy.

What is entropy?

Entropy is a property in thermodynamics that is most simply defined as the measure of disorder. The results of ‘work’ or heat. When objects move by force entropy is created. Heat creates entropy. Entropy is everywhere and it is constantly increasing in the universe. Wikipedia link if you want it.

This is not an absolute definition, it’s also about the probability of the way atoms can organise themselves in different structures. But for the purpose of our sanity, let’s focus on the disorder definition.

Many of us will have inherited and come by product backlogs or products with what I could describe as “high entropy”. Or high disorder. Bugs and stories all over the place. Some detail, some gaps. Epics and spikes scattered like ruins of ancient cities. Not really ranked. This is disorder. Or high entropy.

The more work done, the more heat, the more entropy. Bits of the universe get disturbed by all your hard work. Pushing atoms around. Atoms make things. It’s not a huge leap to believe atoms make Jira items (they do). It’s proper graft. Hence the mess. Atoms all over the place.

Yes, I have debatably managed to link Jira to Thermodynamics. An integration no one asked for.

So in this vibrant, energetic world, the only way to maintain order (in your product and product backlog) is to constantly manage its entropy.

This is the lesson. That entropy is an ever expanding property of nature. It is everywhere, all of the time. You can’t stop it.

We don’t live in a “twilight world”, we will live in an entropy based world. At both dawn and dusk. Disorder and the probability of disorder as far as you can see.

Entropy and information

People at IBM in the 1960s. Before data centres looked like black steel fridge festivals.

The whole information-entropy thing was so confusing that in 1960s a leading scientist from IBM research, Rolf Landauer, got involved to prove that storing and erasing information even creates entropy. They were trying to confront an old thought experiment called Maxwell’s Demon.

They basically worked out that even deleting things, creates entropy.

So we are not the only people to have gone down this path. If it was good enough for old big blue, let’s continue ourselves.

Want more on information entropy? Read Landauer’s principle or about the heat created when a qubit is erased.

Product entropy?

It’s therefore pretty easy to believe that (digital) Products themselves have entropy. The must do if information does.

Digital products use information to deliver value to us as users.

Want an example? Take the massive volume of content moderation done by both humans and AI on social media platforms.

Content disorder and user generated entropy on Facebook

These content moderation systems exist because of disorder and the probability of it.

Take a feature, a comment thread on Facebook or Instagram, delivered at scale results in massive movement. Millions of users typing and tapping. Facebook users tapping their device screens, heated up by lithium batteries.

Data centres and servers spin up, scaling with their spinning cpu fans. Power plants run their turbines to meet the demand. Heat and work as physics would define it. Results in the probability of disorder (high entropy). That results in a problem. That results in an opportunity. That results in a business. And so the cycle continues.

So, it maybe a stretch, but on that basis Entropy could even be optimistically defined as a foundation of innovation.

Entropy has created the opportunities for products and services that help mitigate or reduce, disorder.

By accepting entropy, by budgeting for it and confronting it, or exploiting it, you can manage it.

You can even turn it to your advantage.

If entropy is disorder, then it is chaotic and as they say on Game of Thrones “chaos is a ladder”. One organisations chaos is another organisations opportunity. Now, this is product management and this is familiar business ground.

An example of entropy as a tool?

Netflix are a great example of a company that embraces entropy. Let’s take a quick look at their chaos monkey testing approach. They purposefully introduce entropy, simulate chaos to see what breaks and how. Then proactively fix things that a big dose of entropy has exposed, before real world entropy has had a chance to bring its own chaos.

Netflix engineers code simulated chaos(high entropy) to predict a possible future, then improve things to reduce the probability of that future occurring.

Netflix and don’t chill, panic cos production is broken

Entropy in Product Management

The entire product management journey is about both fighting entropy and embracing it. Managing the universe’s natural desire to become more disorderly. More chaotic. More vibrant.

Products themselves, through usage, the creation and caching of data, the addition of features, the entire lifecycle of products and product delivery creates entropy at every turn.

Backlog entropy

For example a small backlog, ranked and ordered has low entropy. That is good. High clarity, low chaos.

A large backlog, with lots of items, unordered, has high entropy. That is bad.

Your usual window to Product entropy

BTW, well done if you got this far. Don’t worry, we’re finishing up.

What did we learn?

So, after all that, we could summarise:

  • Entropy is a result of work
  • High entropy is disorder
  • Disorder creates opportunities
  • Keep things simple to reduce the opportunity for entropy
  • Entropy applies to information
  • You can simulate disorder to build more robust systems
  • You can’t reverse, delete or stop entropy increasing

So maybe the best advice is to embrace this property of thermodynamics, embrace chaos and disorder. Use the opportunities that arise from it.

By all means strive for a low entropy product backlog, or a product low in entropy. But work in the knowledge that the laws of probability and the universe itself is against you in this endeavour.

Introduce entropy to make your product more robust.

But most of all just accept it. Because it is universal.

You can’t even delete it

Thanks for reading. 🙂

I’ll leave you with two product managers discussing their entropy based OKR