Since the arrival of the Covid-19 pandemic, there has been a huge upsurge in the demand for what is being called “fast science”. Groups around the world are clamoring for data and information on the SARS-CoV-2 virus that causes the disease; not surprisingly they want it as fast as possible. This has led to unprecedented sharing and collaboration between universities, institutes, governments and industry[1]. Whilst tools to share data and information, and to interact virtually with potential collaborators are already available, there still remain issues around how to make fast science work effectively[2]. In this blog I want to remind people that speed is not the only measure; fast science is no good if the quality is not there. Consequently, it is vitally important that “fast science” is also taken to mean, implicitly, fast AND GOOD science.
Scientific discovery can be seen as the iterative process of Hypothesis, leading to Experiment, followed by Analysis of data and observations, then Sharing of results and formulation of a new hypothesis, leading to new experiment, etc. (Figure 1). Ultimately this discovery cycle is broken when a novel insight, backed up by appropriate experimentation and analysis, is deemed worthy of even wider sharing with peer review and publication. Simplistically, “fast science” is this same iterative cycle, but done at speed.
Figure 1
To truly enable fast science, the question now becomes: which parts of this cycle are amenable to speeding up such that the whole cycle is faster? At some level they can all be accelerated using appropriate collaboration and technology tools[3]. Hypothesis and idea generation can be achieved faster with more social, free-form tools[4]. In the experiment space, effective sharing and rapid evolution of methods and protocols[5], plus the rapid dissemination of data would seem to be the best ways to speed up this part[6]. Fast collaborative analysis is perhaps harder to achieve, but again, many of the modern electronic lab notebooks (ELNs)[7], combined with suitable visualization tools[8] can enable effective collaborative analysis[9] albeit perhaps not always “fast”. Sharing of data and results can be accomplished quickly using traditional methods of data and document sharing whilst the discovery iteration is in train, but the final publication of a suitably peer-reviewed complete piece of work or study can take some considerable time[10].
But how can scientific discovery be speeded up while still retaining high quality?
The simplest and by far the best way to achieve this comes about through trust. If two scientists trust each other and the quality of their work, then rapid, real time, interactional collaboration can happen relatively easily. But trust is not quickly or easily established, and can disappear in a moment. Nevertheless, I believe that the three axes of Speed, Quality and Trust, as opposed to the so-called “iron triangle” of project management encompassing Quality, Cost and Time[11] are critical to enabling fast and good science (Figure 2).
Figure 2
If you have low trust, then to ensure high quality, two collaborators have to take more time to be sure that the methods, data and conclusions are solid. This makes the collaboration more asynchronous, more “transactional” and slower. This is unavoidable if both collaboration parties want to retain their reputation. One can extend this to accommodate a number of different fast science versus slow science scenarios:-
High Speed + High Quality = (depends on) High Trust
Low Trust = Low Speed (regardless of Quality)
Low Quality + High Speed = (leads to) Very Bad Science + Very Low Trust
Low Quality = (leads to) Low Trust
High Quality + High Trust = (enables) Good, Fast Science
Whilst this is not a complete set, it serves to emphasize the point. The challenge then becomes one of how to build scientific trust quickly? That question is not easily answered or solved. However, I would maintain that trust in a collaborator, partner or work colleague comes from both the objective – reproducing their work and successfully getting the same results – and the subjective – sitting firmly on the softer side of relationship building and maintaining. Neither approach, objective or subjective, happens quickly. So, whilst good, fast science is a goal we are all trying to achieve in these extraordinary times, it can only really be achieved if:-
1. We already have high trust in the partners and collaborators we want or need to work with. This is a lower risk approach but is not by any means guaranteed to lead to success.
2. We are prepared to give collaborators the “benefit of the doubt” that they do good work and is most likely to be based on their reputation (however that might be measured). This is a higher risk approach, but is one which will likely predominate through this current Covid-19 crisis. On the plus side however, if such “benefit of the doubt” collaborations do bear fruit, there will be many more relationships of type #1 established, which can only help good, fast science become more likely in the future.
In conclusion, fast and good science is dependent on trust, with reputation playing a significant role too[12]. It will be interesting not only as part of the global response to this current Covid-19 pandemic, but also in the world’s response to future acute, critical and potentially existential crises (e.g. another pandemic) to see how science and collaboration has learnt and evolved to be more responsive and fast-acting. But with trust remaining an absolutely critical component, it will be the collaborative behaviors not just of scientists themselves, but also of the organizations, institutes and companies they work for, as well as of the countries where they live, the governments (who often provide their funding) and their leaders, to foster inter-organization and inter-national working which will become ever more crucial. Meantime, we hope that good, fast science will grow so that the light at the end of the Covid-19 tunnel is reached sooner rather than later.
References
* Image of Roadrunner courtesy of https://sketchok.com/cartoon-characters/various/how-to-draw-road-runner; Please note: All hyperlinks were active at time of writing, May 2-5, 2020
1. https://www.covid19dataportal.org/; http://www.qub.ac.uk/coronavirus/analysis-commentary/collaborative-research-culture-needed-to-address-covid-19/; http://www.pharmatimes.com/news/nice_joins_international_covid-19_pandemic_response_1339372
2. Sarewitz, Daniel. “Slow Science, Fast Science.” Issues in Science and Technology 36, no. 3 (Spring 2020): 18–19. https://issues.org/slow-science-fast-science/
3. Please note, in this blog I am taking “collaboration” to be a broad concept; so not only within a team or a group within one institution, but also between groups or organisations.
4. Classic videoconferencing tools like Zoom, Skype and Teams can help here as can platforms such as Slack, and newer technologies like Mural (https://www.mural.co/) and Miro (https://miro.com/). See also: https://biz30.timedoctor.com/online-collaboration-tools/
5. ELN technology and systems such as Protocols.io (https://www.protocols.io/) can play a major role here.
6. See for example: https://www.who.int/bulletin/volumes/98/3/20-251561/en/
7. Kwok, Roberta. (2018) Nature 560, 269-270. DOI: 10.1038/d41586-018-05895-3. https://www.nature.com/articles/d41586-018-05895-3; https://en.wikipedia.org/wiki/List_of_electronic_laboratory_notebook_software_packages
8. https://www.ngdata.com/top-tools-for-data-scientists/
9. Isenberg, Petra & Elmqvist, Niklas & Scholtz, Jean & Cernea, Daniel & Ma, Kwan-Liu & Hagen, Hans. (2011). “Collaborative Visualization: Definition, Challenges, and Research Agenda.” Information Visualization. 10. 310-326. DOI: 10.1177/1473871611412817.
10. Kelner, Katrina. (2007). Science. DOI: 10.1126/science.caredit.a0700046. https://www.sciencemag.org/careers/2007/04/tips-publishing-scientific-journals#
11. Atkinson, R. (1999). “Project Management: Cost, Time and Quality, Two Best Guesses and a Phenomenon, It’s Time to Accept Other Success Criteria.” International Journal of Project Management, 17(6), 337–342. doi:10.1016/s0263-7863(98)00069-6
12. I have discussed trust in science, reputation and reproducibility in previous blogs: e.g. https://www.linkedin.com/pulse/blockchain-distributed-ledger-technology-enhancing-trust-shute/