The return of Agile Methods for research

June 21st, 2019 / By Ryan Butterfield, DrPH, MBA, Paul LaBrec

This blog is an add-on, addendum, follow-up, sequel, or closing score to our previous discussion on the use of Agile Methods in a research environment. We have been applying agile techniques to several of our research and product development projects over the past couple of years, and can share some thoughts from our experiences and walk through some tips and techniques based on our recent project adventures.

For a simple thought experiment, how would you go about QA-ing the production of nanocytes? These remarkable microscopic machines have very defined roles, so how would you recognize malfunctions? Even if they are working, but not in a reliable and valid way? If nanocytes are programmed to detect and eliminate cancer cells, or to seal wounds on a futuristic battlefield? What would the R&D process for this even look like? These are the types of questions researchers will soon be facing. Now, what does this have to do with Agile Methods? Perhaps nothing, perhaps everything. Scientific process and method can no longer consist of just throwing darts at the board. It must contain the stages of inquiry and detection and also improvements in robustness and repeatability. Just do a quick web search for “p-value crisis in statistics.” Yeah, there is not a lot of good going on with that topic.

The scientific members of the academic community live in a publish or perish, Game of Thrones-esque world where grants and publications directly equate to success. In the corporate research setting, less emphasis is placed on publication and more on invention and production, causing its own competitive pressure. Competition happens in the marketplace and through internal company pressure to bring new products to market and meet financial goals. There’s no doubt science is a big economic force. We recently attended a conference in Dallas, TX where Dr. Michio Kaku discussed the science of the future and its connection to the economy of the future. If a renowned futurist and quantum physicist evangelizes the potential of science as an economic driver, then perhaps we need to think about how we create scientific processes which enable robust, repeatable and highly producible products/goods/services for that future.

Here are some potential cons of Scrum/Agile Methods based on a few of our recent experiences, validated through this short poll:

True or False questions to ponder:

  • Agile Methods Work? (T F),
  • Agile Methods increase stress on employees? (T F),
  • Shorter deadlines are always better? (T F),
  • Agile Methods increase micromanagement, therefore employees have less freedom? (T F),
  • There is a general increase in productivity when using Agile Methods? (T F),
  • The more black belts you have, the more boards you can break with your foot? (T F)

We jest with these questions…slightly. We pointed out in our previous blog [ADD LINK] that while Agile Methods are great for things like a programming project or a QA review, when it comes to research, the most valuable assets are people and time. So…the question on the table is, “How do we optimize time so that our people can be as creative and efficient as possible?” The last thing you want to do is soak up the available time of your talent pool in wasteful meetings and misdirected lines of inquiry. How to be efficient? One advantage to using Agile Methods is the adoption of an agile mentality among our teams. Creating a standardized approach to a research project that each of us follows (regardless of the project or team member) allows a certain flexibility that wouldn’t be possible with the rigid, all hands on deck, time-consuming approach that is so prevalent in R&D settings. Another potential benefit to having a universal Scrum mentality is that all team members can be familiar with each part of a project. which means the whole project isn’t out of commission if the point person goes on a two-week vacation to a small European country with limited internet access. We have found interchangeable personnel with comparative skill sets is a key to success.

Here are a few high-level and useful tips and techniques from some of our recent projects: 

  1. Create a master calendar that is tied to project goals and deliverables.
    • Team members need to hold each other accountable, not just management directing the work.
    • Project plans needs to be clearly set out with linear goals and objectives.
  2. Have centralized project folders with all documentation available to each team member. This empowers the full team and helps reduce reliance on a few key people.

  3. Create reusable research datasets so you don’t expend unnecessary time assembling and validating datasets. Choose datasets that are:
    • validated,
    • “fit for purpose,”
    • known intimately by the team.
  4. Work in iterative steps.
    • Shorter iterations with circumscribed research questions allow you to identify “dead-ends” sooner and change direction.
  5. Target incremental improvement in product releases.
    • Research is never really completed as every answer brings new questions; in our world we are all Pareto disciples.
    • Product releases must be scheduled, however, so the research supporting them must have end dates.

The use of Agile Methods in scientific R&D is very useful when used correctly. It is not a catch-all to create a scientific “factory” to achieve an endless number of grants and published journal articles. Nor do Agile Methods ensure new product versions every quarter. Applying Agile Methods to research projects, however, does result in a process that enables robust, high-quality work that can be performed in an efficient and repeatable manner.

Ryan Butterfield, DrPH, MBA is a statistician/population health researcher with 3M Health Care Business Group.

Paul LaBrec is research director for 3M Health Care Business Group

References on project optimization:

Additional Readings: