Tuesday, 12 February 2013

Operational Excellence vs Technical Perfection

In the modern world of software development, where focus on quality is just as (if not more) important as quantity, it is becoming increasingly important and difficult to find the correct balance between the two.

As companies begin to work in a more 'agile' and lean fashion, adopting techniques such as TDD and pairing, the type of person that they tend to chase in terms of recruitment is also changing. Developers with good skills and experience in TDD and XP are in very high demand as a result of this. The danger here is that a lot of developers with this skill set are very opinionated and have very strong feelings with regards to how software should be written. It is difficult to keep some of the more purist of mind happy, as they want to do everything in a perfect way. The majority of what they say is correct, but it is not always practical to put it into effect. These people have great ideas, and want to have a positive impact on day to day work - this is something we should actively encourage and not stifle. Not getting their ideas into reality can often lead to them feeling under utilised or under valued.

This leads us to the challenge of living in the real world. Given an infinite amount of time these developers would work towards the perfect solution. This is what I would call a development utopia. It is a myth, a world that developers can only dream of, where everything is tested, readable, etc. In reality there will always be better code, and no matter who has written something or how it has been written, the likelihood is that there will always be ways to improve and refactor the code by the next person to see it. The outcome here is that tasks can take a very long time to complete - meaning a much slower time to market for delivering value.

The business stakeholders want to release as many features to the customers as possible. The business value utopia would be quite the opposite of the developer one, with a total disregard for quality of code or the future and only a short sighted what can we get now mentality. They don't necessarily care how we do something, but that we get it done and delivered.

When working in a commercial environment, the overall aim of the team should be to deliver value. However we should all be clever enough to learn from our mistakes that we should not sacrifice the future in order to get more now. This is where we need to work more collaboratively with the business in order to find some kind of balanced middle ground. Yes, we should invest in our software to ensure that we can still make changes in the future, but do we really need to engineer the perfect solution every time? By the same measure, the business need to accept that we can not just churn out tickets constantly if we want to ensure future development capability.

There will always be times where quality needs to be sacrificed in the short term in order to deliver critical business functionality or issues, and as developers we need to be flexible to any such demands. The counter side to this is that the business should not abuse this way of working and should have a very strict way in which to identify any such work. Additionally, the business should be bought in to the way in which technical debt works - and give development teams time to tidy up after themselves should the above ever occur. I have seen first hand where team's will react in a negative manner when such an event happens. Rather than reacting in this way we should be positive about doing our best to help the company as a whole - if we want the business to allow us to work how want to, we also need to be receptive to their needs.

The bottom line is that we are all at work to deliver things that the business wants. Delivering the perfect solution but taking 2 years to do so does not represent value. Delivering 100 things in a week but leaving the code base in an unworkable state going forward might get some immediate gains but would ultimately cripple the business in the future. This is why we need to find the balance between operational excellence and technical perfection.

4 comments:

  1. Interesting post. What you're describing seems more like systemic issues rather than individual development issues. The overall system is trying to find balance and you end up with an extreme spectrum of behaviour.

    I've witnessed a lot of pain as result of not putting quality first. Doing the right thing gets tougher as compromises are constantly taken. This approach lasts for so long until we reach a critical mass with a code base which can no longer be hacked into.

    At this point we witness a failure to reassess the situation and we end up back in the 70s where by "the business" demand more "resource" are thrown at a product in order to meet their demands. Things often get worse where by we try to optimise and manage individuals by time sheets and other means, failing to see systemic issues.

    What happens in this situation? Yup, code quality DROPS, cycle time INCREASES and "the business" end up even more frustrated. What's the solution? Often, throw more people at the problem, despite the system screaming at us that it's broken.

    The bottom line should be that we should all strive to be part of the business and not deliver to arbitrary demands. ALL work should be assessed and validated against the same company goal. Spend two years working on a perfect solution? Fine, you just cost us 2 million pounds and made no return on investment.

    ReplyDelete
    Replies
    1. I think you hit the nail on the head there Seb with that last bit "Spend two years working on a perfect solution? Fine, you just cost us 2 million pounds and made no return on investment."

      We need to be assessing the value not just on the features we develop, but the amount of time we spend making something of a certain quality.

      Delete
  2. Just get it f%$£ing done (as best as you can in the time given)!!!

    I've known this for about 8 years now which is why I'm a Rockstar Dev!

    ReplyDelete