Why we mustn't let perfection be the enemy in our fight for automation
In this blog, Shafi Chowdhury previews some highlights and perspectives from his upcoming talk at the PHUSE EMEA Spring SDE on Tuesday 4th May 2021. The SDE will focus on enhancing productivity through automation, and central coding repositories and Shafi will contribute his insights from his decades of experience.
Q&A with Shafi Chowdhury
What perspectives are you planning to share?
Since I started programming 25 years ago, management has always aspired to the idea of generating outputs at the push of a button. Nowadays, using a central repository for data and harnessing automation has become a hot area for discussion. However, we must remember that these approaches have been discussed and used for years in some form or other. We aren't yet realising the potential of the technologies we have available for various reasons. Certainly, there are several pitfalls and issues to consider when embarking on a central repository and automation effort, and that is what I will discuss in my presentation. There are typically two approaches that are being used alongside a central repository to leverage automation.
One approach is calling on macros, but we must be careful not to let this become like a black box where we don't know how the work has been done. Simplicity tends to be best - I often see macros trying to take care of too many scenarios.
We also have the approach of using template programs that can typically tackle up to 80% of the work on a given project. Programmers usually need to make a few adjustments to make these programs work for their own study.
At any rate, central repository approaches rely on people following good programming practice. If you're looking at a program in six months time, can you understand it?
Where do you feel companies can make improvements in their approach to automation?
From my perspective, automation should deliver the end result in the form of outputs– not a halfway house. I've seen various automation approaches that don't provide that end product; they take you to 80% of the way and then programmers have to adapt their approach. To me, this simply creates more frustration and confusion. We can build towards more complex data transformations, but at this point, for tables, listings, and figures, it makes sense to fully automate. This will ensure consistency and quality, save rework, and ensure that the programs sent, for example, to the FDA or other regulatory authorities, are easy to read and have a complete audit trail.
What would you like attendees at your presentation to take away?
We need to be very clear about what automation is trying to deliver for our organisation. Further, we shouldn't let perfection be the enemy of the good in our drive to realise the benefits of automation in programming activities now. If we can improve 80% of scenarios, this will deliver great efficiencies versus the status quo.
Join the PhUSE webinar here or contact firstname.lastname@example.org to arrange an exploratory discussion about your central repository and automation requirements.