Roadmapping research


Roadmapping research

What is a scientific roadmap?

If you’re studying a STEM discipline, you could use your thesis to create a scientific or technological roadmap. Creating a roadmap  can help identify novel strategies for achieving a goal and prioritise between them, by setting out constraints on the way to a goal and potential workarounds for them.

One approach to creating a scientific roadmap, described in the paper Architecting Discovery, involves:

  • identifying a goal that science and technology has not yet achieved.
  • mapping ‘currently practised and conceived approaches’ to trying to achieve the goal.
  • identifying the fundamental limitations imposed by the laws of physics on how effective these approaches can be.
  • identifying why it seems these fundamental limits haven’t been reached – i.e. what are the current technological bottlenecks preventing this?
  • exploring whether workarounds to these obstacles can be found.

Some potential methods of finding workarounds are:

  • finding a hidden gem that already implements the workaround, by searching through existing knowledge in other scientific domains (or consulting experts in these domains) and transplanting this knowledge to the new domain, or by screening libraries of elements from nature and finding an example of a workaround occurring naturally. 
  • the suggestion from this paper to ‘look at what people are doing and then imagine what it would mean to do the most opposite thing you can think of…for example, for expansion microscopy, if everyone else is zooming in to see a biological system better, what happens if we blow up the biological system (in an even way of course) instead?’

The paper Physical Principles for Scalable Neural Recording is an example of a similar approach to roadmapping. This paper aims to identify promising strategies for ‘simultaneously measuring the activities of all neurons in a mammalian brain at millisecond resolution’ without damaging the brain.

Here are the steps of scientific roadmapping in more detail, based on the paper above:

  1. Establish the goal, e.g. ‘simultaneously measuring the activities of all neurons in a mammalian brain at millisecond resolution without damaging the brain.’
  2. Map out the landscape of modalities (current and conceived) aimed at achieving the goal (in this case, this includes approaches such as magnetic resonance imaging and ultrasound).
  3. Identify how to assess how close each modality is to achieving the goal. (For example in the case of MRI, this is primarily determined by the spatial and temporal resolution that can currently be achieved).
  4. What fundamental limits is the use of each modality subject to? (For example, the speed that sound travels in the brain, or the degree of volume displacement that can occur in the brain without causing significant damage.)
  5. What is the potential of each of these modalities? Do some look more promising than others, in terms of how close they are to being viable?
  6. What breakthroughs would be needed to achieve the goal using each modality? Are there workarounds that would allow you to bypass the physical constraints they are subject to?

What makes a good roadmap of a technical area? 

Some things to bear in are:

  1. In the goal you’re focused on achieving well-defined and quantifiable? 
  2. Is the goal interesting and relevant to some important outcome?
  3. Do you have (or can you acquire) enough understanding of how the system works? This is important so you can draw non-obvious, definitive conclusions from backchaining from the goal.
  4. How thorough/convincing is your backchaining from the goal?
    1. Is the goal decomposable into necessary or sufficient conditions?
    2. Have logical, quantitative, or probabilistic limits been put on different paths to the goal?
  5. How exhaustively are possible routes to the goal explored?
    1. How close to MECE (mutually exclusive, collectively exhaustive) are the routes you’ve mapped out?

Further reading

If you want to learn more about roadmapping research, you could listen to this episode of the Ideas Machine podcast with Adam Marblestone, one of the authors of the paper above. You could also look at his presentation on ‘Roadmapping Biology.’ The paper Architecting Discovery also provides an overview of this type of research in more detail.

Some examples of roadmapping in research papers are: