🎙️ Podcast Link 🎙️

In any research career where you are responsible for conducting or leading a program of innovative research, your success will depend on your ability to generate good research ideas – a process known as ‘ideation’. In your career, you may have already met or worked with researchers who seem to have a near infinite source of exciting ideas for research projects, or can come up with endless new ideas on the spot. But behind those impressive feats is typically a well developed skillset in research ideation. In this video I dive into the ins and outs of becoming effective at research ideation, and the benefits it can have for your career.

I touch on key concepts, like the fact that much ideation is actually a continuous, spontaneous process, where ideas can pop up at any time, and where having a system to note and later follow up on those ideas is crucial. I talk about the various ways in which you can think about whether an early stage idea you have is potentially a good one – including aspects like potential benefit, specificity of benefit, and some meaningful additional contribution over what has already been done in the field. I give examples of ideation in a computer science context, and talk about the different stimuli you can use to drive ideation, from research methodology to end user problems. When ideation is done with a specific purpose in mind – for example for an academic paper or grant submission – there are extra factors to consider, like topicality, fit for your profile and interest level. Finally I highlight how being good at ideation is a universally useful skill, across a number of research-orientated career types, and for both early and late in your career.

Check it out, in this post, on YouTube, or via podcast.

Please reshare if useful 🙏

🕒 Timestamps are as follows:

📌 (0:00) Introduction to Ideation in a Research Context
📌 (0:53) Research Ideation is an Ongoing Process
📌 (1:27) Ideas Can Pop Up at Anytime
📌 (1:51) Jotting Down Ideas For Later Consideration
📌 (2:39) Consider a More Detailed Ideas Log As Well
📌 (2:51) What Makes For a Good Idea?
📌 (3:02) The Need for Some Type of Benefit
📌 (3:11) Three Benefit Examples in a Computer Science Context
📌 (3:49) Separate Benefit Articulation from Chance of Success
📌 (4:05) Benefit Can Be Immediate and Stand Alone or Long Term and Integrative
📌 (4:18) Being Able to Articulate the Potential Benefit is Key
📌 (4:35) Specificity Can Help Refine Your Ideas
📌 (5:07) Specificity Example: Computational Advantage
📌 (6:23) Benefits are Very Rarely Evenly Spread
📌 (6:45) Ideas and Novelty
📌 (7:14) Not Everything Has to be Novel!
📌 (7:26) When Learning to Ideate, Don’t Obsess Over Novelty
📌 (8:08) Ideation Stimuli: Approach Versus End-User Problem
📌 (8:24) Research Approach Example: A New Machine Learning Technique
📌 (8:35) End-User Example Problem
📌 (8:54) Choosing Between an Approach versus Problem Perspective
📌 (9:12) Sharing Ideas with Peers: Pros and Cons
📌 (9:38) Ideation in Your Area of Expertise is Usually More Efficient
📌 (10:22) Ideation Driven by Areas of Interest
📌 (10:40) Remember Interest and Capability Don’t Overlap Exactly
📌 (10:55) Ideation with a Specific Outcome in Mind
📌 (11:13) Balancing Opportunity with Your Capability Fit
📌 (11:50) Ideation is Useful for Many Career Types and Stages
📌 (12:19) Having Some Ideas Helps Generate More Ideas!
📌 (12:34) The Joy and Excitement of Ideation

Full Video Notes

• One of the key capabilities for a successful research career is ideation. In this context, ideation is the process of coming up with ideas for research. Prolific researchers are often excellent at ideation – which means that they are able to generate lots of ideas for research without a lot of effort. Being good at ideation also means you have a good sense for what makes for a good research idea in your field, meaning you are able to rapidly filter and focus on the most promising ideas you come up with. This capability in turn means that experienced ideators are never short of ideas for their own or their teams’ research efforts. They’re also never short of ideas for a new academic paper or a grant proposal. In this video I’m going to cover some key concepts and tips and tricks for becoming effective at research ideation. 
• The first key concept to grasp is that research ideation is not a rigidly segmented activity in your day. Sitting down for an hour and “deciding to ideate” without any preparation or prior context is not a recipe for success. When you encounter a researcher who is overflowing with ideas, it’s often because they are cultivating a huge list of ideas over time. These researchers are always generating new ideas, sometimes at the most unexpected times.
• Ideas can come to you just before bed, when you’re out and about, on a run, or when you’re in the shower. My area of research for example is in navigation and positioning systems for robots and autonomous vehicles. The nature of that topic means that ideas can come to me whenever I’m out and about and moving through the world, whether driving, running, cycling, flying – whatever. Consequently, it’s important to have a system whereby you can jot down these ideas as quickly and effortlessly as possible. This is especially if it’s important that you don’t interrupt whatever activity you’re engaged in – for example during your precious time with family or friends. One solution to this is some sort of note taking app on a smartphone, given many people almost always have one with them. Another is the traditional pen and paper, or biro and napkin, depending on what materials you have to hand. You only need to jot down a few words – and it can be fun to come back to what you’ve noted later and attempt to decipher your cryptic wording. And it is important that you do come back and revisit your idea in a little more detail. You can do this with an online spreadsheet or document, or a more detailed ideas journal.
• So what makes for the beginnings of a good idea? The exact parameters vary a little by field but there are some commonalities.  
• Firstly, the idea you have must have some promise in terms of a potential benefit. What that benefit is can vary hugely. In computer science fields like robotics, machine learning or computer vision, one type of benefit might be an algorithm or machine learning technique that requires far less computational power, saving on hardware cost and reducing energy consumption. Another type of benefit might be some new theory that provides provable or bounded performance guarantees for a system. A third type of benefit could be a computational system that has some sort of introspective capability – the ability to “know when it doesn’t know” for example.  
• When thinking about benefit, it’s vital to separate the chances of your research idea actually being successful from the benefit it would have, if successful. Most of the time, no-one is looking for guarantees that research will be successful. The benefit can be immediate and independent from other scientific developments, or it can be a benefit that would be realized in 30 years time, and only with commesurate advances in related fields. What is vitally important is that you can articulate a clear and concrete vision of how that idea, if successful, would have some tangible benefit. A lot of ideation sessions come unstuck because researchers are not able to articulate these benefits. 
• As you develop some promising ideas, specificity is critical. Inexperienced researchers will often start off with a blanket broad benefit statement about the potential benefits of an idea they have, but trying to article those benefits as specifically as you can can help shape and refine that idea. It will be a very rare idea indeed that makes major advances in many ways – a typical good idea will advance the field in one, or perhaps a couple, of very specific ways. 
• To illustrate this concept of specificity in a computer science context, imagine you’ve got an idea for a new technique for object recognition for a robot or autonomous vehicle, and you think that there is a computational advantage to that approach. What exactly is that potential computational advantage? Does it require less computational power to train the system? Less computational power to deploy the system? Does it cross a critical absolute threshold whereby real-time computation on a compute-limited small robot or drone becomes possible, rather than needing offline, offboard computation? Is it so computationally efficient that you can train hundreds or thousands of these systems in parallel at ten times or hundred times real-time speed? Does it run only on a CPU, rather than requiring specialist compute hardware like a GPU? Is its strong spot equivalent or near-equivalent task performance to the current state of the art with far less compute requirement? Or does it, when scaled up to run on equivalent compute to the current state of the art, achieve better performance? To be clear, you’re not trying to guarantee any of these outcomes, but to engage in the process of thinking where the specific benefit might be. It is very rarely an evenly distributed benefit across all of these areas, and thinking about where the idea might have maximum benefit is a very useful way to shape its further refinement.
• Another key property of good ideas is that they are ideas that have not already been entirely fleshed out and investigated to their full extent. This is often referred to as novelty and the takes on this vary hugely between some fields. The minimum threshold here is simply that your idea, if it pans out, would add some meaningful additional contribution to the scientific, societal, industry or other communities. That does not mean that an algorithm you propose has to be entirely novel – perhaps it’s the way you’re proposing to use it that has some novelty and hence some potential new benefit over the status quo. 
• For researchers who are relatively new to ideation, a word of caution around this concept of novelty. A lot of inexperienced researchers spend too much time agonizing around whether the specific idea they’re thinking of is novel. But at the early stages of your ideation journey, it’s much more important to practise coming up with compelling ideas, whether they are novel or not. Adding the extra criteria of sufficient novelty, whatever that means for your field, is going to get in the way early in your ideation learning journey. This is also where working with, and watching experienced ideators can also help. 
• Different people drive their ideation in different ways. Two of the most common are ideation through thinking about research techniques or methodologies, and ideation through thinking about end-user challenges and problems.
• An example of the former would be thinking about a new machine learning architecture and brainstorming interesting ways to use it, given its strengths and weaknesses. An example of the latter would be distilling a key problem theme that the industry or government partners you’re talking to are telling you about. An example of this in my field might be the assertion that autonomous vehicle technology is too expensive to be feasible. Ideally you would consider both technical approaches and end-user problems in your ideation sessions. In the early stages of ideation though, most people find one or the other of these approaches to be much more natural, and hence more prolific in terms of generating ideas. 
• Should you share your early stage ideas with peers and colleagues? The default advice here is a resounding yes – you will gain so much by testing your ideas with others and getting their input as well. The main exceptions here would be if you work in an environment or field where ideas can be very easily scooped, and if you don’t have that level of trust that someone won’t steal your idea. 
• You will generally find it easier to generate compelling ideas in your field of deep expertise. That doesn’t mean you can’t stray outside of that field, but a lot of the bulk of idea generation will typically occur close to what you know. The average viability of these ideas will also be higher, because you understand the nature and expectations of your field. Idea generation in fields you’re not intimately familiar with will often result in further due diligence on the idea rapidly finding out its limitations, or that it’s already been done before. The advantage of ideation in areas you’re deeply familiar with is you’ll often know these limitations or prior work off the top of your head.
• Idea generation is also easier in fields and areas that you are excited about and interested in. Ideally your ideation will primarily occur mostly at the intersection of the topics that you’re most interested in, and the topics that you are most competent and experienced at. It’s important to note that these are overlapping but not completely overlapping areas – there will be areas that excite you but which you’re not very experienced at, and you will often be good at some things that don’t excite you that much. 
• If your ideation is somewhat targeted towards concrete outcomes – for example towards generating ideas for academic papers or grant proposals, you should also consider the level of opportunity in different topic areas, and your level of track record and experience in those areas. Especially for grant opportunities, the return on investment will rely on some combination of the amount of opportunity for funding in those areas, combined with your fit and profile in those areas. In some situations you may consider ideation in a field that doesn’t get a lot of funding, but where you are clearly one of the world leaders for research in that field. Conversely, you may ideate in a field that is overflowing with funding because it is very topical, without necessarily having to be a world leader in the area. 
• Ideation is a skill that, when well-developed, will serve you well in many types of careers and at all career stages. It’s useful when you are a sole contributor working on your own research, but also when you are leading a large team and helping them generate ideas. If you can become prolific at ideation it can also facilitate a network of collaborators, where you can share your ideas and work with others to realise those ideas. It’s also a skill that builds on itself very well. People who have practised generating lots of ideas, can then use those ideas to generate yet more ideas, more easily than someone who is starting from scratch. It can also be a joyous experience – there’s nothing quite like that dawning realisation that you may have thought up something really transformative. Most ideas of course aren’t like this – hence the need to become prolific at generating them. But when you do come up with that cracker of an idea, it can be incredibly exciting!