Learn to remove – it may be harder, but the results are way better | Working Principles

I’ve worked around corporate IT systems for most of my adult life and what follows is a common, if slightly embellished, history of how organisations get into a mess:

• We had a need for a system to do X.
• Another team had a need to do Y. The managers of team X and Y are in different parts of the organisation that don’t like to collaborate.
• Despite the lack of collaboration we decided that we needed some of the information from Y in X. We built another system, Z to move some of the data between them.
• Y didn’t work very well so we built A, but never decommissioned Y. We had already added some customers to Y and it is always difficult to move customers. One customer who uses Y is particularly difficult.
• People liked to use A so we added some customers to it, but different customers from the ones that use Y.
• Another team in another part of the organisation build B. We then discovered that B was similar to Y but worked differently, so we built C to make them work together in a similar way to Z.
• We then got a new leader who had worked with D in their previous organisation and have spent the last year trying to get D to work like Y, with data from X, A and B. We needed another system, E, to move the data between D and Y as an interim solution while we did the development.
• The leader who was a fan of D has since left the organisation and everyone is unsure of its future. The technical people the leader recruited from their previous organisation have also left to join them at their new venture.
• The team that built Y has since been allocated to other work so no-one in the team knows how it works anymore. We need Y to work, because the difficult customer is still using it.
• We’ve recently experienced problems where we’ve been getting inconsistent results from some of these systems and it’s becoming embarrassing with our customer.
• Given these recent issue there’s a strong tendency for action hanging over everyone and the prefer actions is to add something. “Perhaps we need system F?” says someone “I’ve used it before and it was brilliant.”
• The reality is, though, system D will do everything every team needs and D is already being paid for, but moving everyone over to it will take work. It’s going to be particularly difficult to move the long-term middle managers over who are heavily invested in X, Y, and Z. Each team is convinced that what they do, and how they do it, is unique and vital to the running of the business.
• No one really knows how A, B and C work and are scared of touching them, fearing a catastrophic breakage. That fear includes a fear of shutting them down.
• Another challenge is going to be moving the customers away from Y and A, what’s in it for them? There’s also that difficult customer to worry about.
• Perhaps adding in F isn’t such a bad idea after all?

One addition has lead to another. The result has created further complication and even more technical debt.

There were plans to remove some of the technology, but they were never achieved. The whole thing has become like that tangled box of cables you have stored away somewhere.

I’ve seen the same thing with processes with particularly experience of review processes. The story is almost identical to the above:

• Review A spawns review B and C.
• Review B and C spawn reviews D, E, F, G and H.
• Review H spawns review I and J.

Each of these reviews takes an hour and 5 people (if you are fortunate) – the burden of a simple review is 50 hours without preparation time.

There’s always going to be a noisy middle manager who insists on having their own review meeting and it’s not always easy to resist these demands, but giving in to them has a very high cost.

Before you know it people are spending more time reviewing things than creating things, particularly the poor people presenting.

It is hard to remove in these situations, but the results are way better than adding “system F”. You know that taking everything out of that cable box and throwing half of it in the recycling is a much better answer than trying to wheedle out the one cable that you are looking for. You know that shutting down the spawned reviews is a much better answer than keeping them going and frustrating everyone.

Adding “system F” might feel like a better answer, but it’s a bit like eating another cookie to help you feel better about being unfit.

Header Image: This is St. Mary’s Church, Longsleddale looking down the dale on a recent spring morning.