Our principles are rules of thumb that work for our organization. We learned them by spotting patterns in what leads to success and what doesn’t. These evolving values, customs and processes are an expression of the culture that defines the London Institute.
Let researchers focus entirely on research.
There’s a bizarre inefficiency in the way we organize scientific research. We identify the best scientists, then impose fierce limits on how much science they can do. The explanation for this strange state of affairs is that the overwhelming majority of science is done at universities, where scientists are expected to devote most of their energies to teaching and university life. What you’re left with is legions of talented researchers, who feel lucky if they get to spend a day a week on research. That’s why we founded the London Institute for Mathematical Sciences. We wanted it to be a place where scientists are free to dedicate themselves full-time to science.
Protect curiosity-driven research.
Curiosity-driven research, also known as basic science, is the process of describing patterns. Applied science harnesses them to address fundamental human needs. It sounds obvious to say that patterns must be described before they can be exploited, but this is not always understood. Many governments and funding agencies look for novel solutions to pressing problems without investing in finding the patterns that will solve them. The trouble is, we never know which patterns will crack which problems, so we must continually invest in the expansion of our pattern repertoire. The London Institute is a rare place where curiosity-driven research is the rule, not the exception. The result is more foundational and ambitious research that has the potential to be truly transformational.
Hire researchers who are mathematically fluent.
The universe is rich in patterns, and mathematics is the language for describing them. All of our researchers are fluent in mathematics, so they can understand each other’s problems and be open to collaboration. A culture of mathematical proficiency also increases the consensus about what constitutes good work. One aspect of this consensus is mathematical beauty: a kind of elegance and concision, which tends to go hand-in-hand with truth. Another is what G. H. Hardy calls mathematical seriousness: this can be grasped by the significance of the mathematical ideas the work connects, which tends to go hand-in-hand with importance.
Hire people who care about the Institute.
The London Institute is a unique alternative to the university model for research in the mathematical sciences. Its small size means that every member plays a role in determining its future. We look for members who are curious about the work of their colleagues, keen to attract others more talented than themselves, and care about the Institute’s scientific success and financial stability. They make the Institute super-additive. On the other hand, researchers who are mainly concerned with advancing their own careers make the Institute merely additive.
Show up and interact
The Institute is more than the sum of its parts.
Organizations exist to be more than the sum of their parts. Otherwise, their members would be no better off working together than apart. Yet many academic research centres are merely additive: their members contribute their own discoveries, but do not shape the discoveries of others. Whether or not an organization is super-additive depends on its culture: its values, processes and customs. The distinctive culture of the London Institute, described in these principles, is built around developing an organization that exceeds the sum of its parts.
Work at the Institute.
Unlike some university departments, the London Institute expects researchers as well as non-researchers to be present during working hours. Being present is especially important to us because of our small size. To reach the critical mass of ideas and interactions that leads to a buzz of discovery, we need everyone on board. Being present fosters an innate awareness of the Institute’s activities and values, which is the basis of our strong culture. It allows for the cross-pollination of ideas and increases the likelihood of unplanned, rather than planned, interactions, which tend to be the source of serendipity.
Colleagues who are present are a resource.
People are social animals, who lend each other energy in ways they don’t always appreciate. The awareness of others doing good work in the next room provides a spur to concentration. In the best work cultures, you are inspired by those around you and want to live up to them. Emotions are contagious, and the joy of insight can spread through the building in a virtuous cycle. When you know that your colleagues are present, and are familiar with their skills and interests, you can enlist their help at short notice. Problems that take days to solve at a distance can often be solved in minutes with direct help from the right person.
Think carefully about which problems to tackle.
While working at Bell Labs, information theory pioneer Richard Hamming asked his colleagues to name the most important problems in their fields. Later, he asked them what they were actually working on. To his surprise, there was no overlap between the two sets of answers. Yet as he pointed out, “If you don’t work on important problems, it’s unlikely you’ll do important work.” Importance is not the only reason to work on a problem. Mathematical beauty is another, since beauty often goes hand-in-hand with physical truth. In either case, you should choose your research problems thoughtfully, not by jumping on a bandwagon or pursuing a random walk through idea-space.
Don’t let managing science replace doing it.
The joy of insight is what attracts scientists to research in the first place. But if you spend too much time managing junior researchers instead of doing research yourself, this spur gets lost. You also lose the technical prowess required to be at the forefront of theoretical research. We expect our scientists to stay active in research and get their hands dirty doing calculations and proving theorems. To protect researchers from becoming merely research managers, we limit to four the number of junior scientists they can supervise at any one time.
Maintain a spread of research areas.
The Institute tries to maintain a roughly even distribution of research interests across the mathematical sciences. In practice, this means a majority of our research will be spread over the core theoretical branches of physics—physics being the dominant mathematical science—with the remainder in pure mathematics, machine learning, quantitative biology, mathematical finance and other areas. When we detect a dearth of research in one area, or a surplus in another, we try to redress the balance with subsequent hires.
Hire non-researchers as talented as researchers.
Many university departments invest heavily in hiring talented researchers but settle for second-rate non-researchers. This is a mistake, because a successful research organization requires more than science: it needs developers, designers, writers, fundraisers, financiers and assistants. Just as much as the scientists, they should be exceptional. We work as hard to recruit talented non-researchers as we do talented researchers. The result is a mutual respect between scientists and non-scientists and a more efficient organization in which members can focus on what they do best.
Mix researchers and non-researchers.
When researchers and non-researchers are segregated in different buildings, or different parts of the same building, it is easy for communication to deteriorate, replaced by unsupported negative narratives. Tasks take longer because there is none of the instinctive awareness of purpose that comes from sharing the same space. Our researchers and non-researchers are mixed throughout the building and work closely together to fund, carry out and communicate our research. Everyone is invited to our events, from communication scrums to research seminars.
Avoid academic departments.
For a small organization like the London Institute, academic departments do more harm than good. When a research organization has departments, scientists tend to assume that they have common interests with those inside their department, and that they don’t have common interests with those outside their department. Both assumptions are often false, in part because department structures are slow to reflect emerging disciplines. Departments can also lead to internal politics, especially when people identify with the department more than the organization. At the London Institute, we mix everyone up inside the same building, so ideas and techniques can cross-pollinate.
Simplicity drives innovation.
When a system is complex, making a small change can cause it to react in unpredictable ways. But in a simple system, the effects of small changes tend to be clear. The adjacent possible is the set of ways that a system can be marginally modified. When it’s hard to understand the behaviour of the adjacent possible, we don’t know which modifications are good and which are bad. As a result, stasis becomes the default behaviour, and the pace of innovation slows to a halt. For example, in a complex web page or contract, it’s hard to make small improvements, because any one change can have unexpected knock-on effects. This is one of the reasons why we are relentless in seeking simplicity—in everything from the structure of the organization to graphic design.
Speed is a form of simplicity.
No one likes to wait, whether it’s for a file to download or grant funding to start. But it’s not just the drudgery of waiting that is problematic. Having to wait for things makes life more complicated, and things that happen fast make it simpler. In the same way as interoperability across space drives simplicity, so too does interoperability across time. Nothing is more interoperable than an immediate solution to a problem. But that interoperability diminishes as the wait time grows. For example, if it takes a year to win a grant and start work, the grant may have lost value, since your research will have moved on. Getting things done fast is one of the Institute’s strategic advantages over other research centres—from building a web page to seizing a grant opportunity to hiring for a new role.
The agility quotient is a way to think about speed.
Imagine someone asks you to do a task that takes y units of time, and let x be the number of units of time at work before you actually do it. Your agility quotient is your typical value of x/y. For example, if you’re asked to spend two hours reviewing a paper, and you get it done in a week, your agility quotient is 40/2 = 20. But if you take a month, it’s 87. People at universities tend to have AQs on the order of 100, whereas for people in businesses it’s on the order of 10. Sharpening the AQ of the Institute’s members gives it an exponential advantage over other research centres, especially for projects that involve multiple people.
Our website is how most people perceive us.
The London Institute doesn't have a lot of visitors or an alumni base of students. Our products are our discoveries, which are conceptual rather than concrete, so we don’t have a large user base. To most people, therefore, we are our website. This is why it’s so important that our website communicates the spirit of our organization as well as the content of our work. Our site must be comprehensible to the general reader, while containing more detailed information for experts. The difficulty of this balancing act is one of the reasons we care so much about the quality of our science writing. How we tell our scientific stories online is key to how we are perceived.
Our website helps us perceive ourselves.
Our website also helps us know ourselves. This may sound paradoxical, since we are responsible for creating it. But self-knowledge is hard, for organizations as much as for individuals. The dictum “you are what you do frequently” helps discriminate between intention and reality. It normally is held up to the individual, but it also applies to organizations. This is why an up-to-date, comprehensive overview of what we have achieved—our papers, projects, events, funding and spin-offs—helps us see squarely where we are. By contrasting this with where we would like to be, the right next steps come into sharp relief.
Our website reflects the Institute in real time.
Given the importance of our website, we want it to reflect the Institute in real time. It should be an image of who we are today, not who we were yesterday. To achieve this, everyone involved in the design and development of our site works in the building full-time and is attuned to the Institute’s activities. This agility is reflected in the user experience: we keep the site lightning fast using cutting-edge development techniques such as code splitting, prefetching, lazy-loading and image optimization.
Tweet mainly about scientific discovery.
The London Institute mainly tweets about scientific research, steering clear of social policy and current events. We tweet about our research papers when they are completed, accepted and published; our research grants when we win them and when they start; and our events on announcement and after they take place. We also tweet about new hires, press coverage, research projects and sections of our website, and significant achievements of our members.
Modularize text and images to speed up design.
Our website is organized around the concept of Lego-like building blocks of text and images on a recursively divisible grid. This visual system for combining components helps us quickly build and adapt pages and opens up new opportunities for consistent and beautiful design. We use a restricted set of image sizes to make it easy to snap them into place across different arrangements. More unusually, we also use a restricted set of text block lengths. For example, paper synopses are 140 characters and biographies are 700 characters. Making units of text that do the same kind of job the same length encourages concision, so that the reader gets the message faster. It also harnesses geometry to indicate the purpose of the text. This speeds up perception, because users take in the spatial relation of information faster than the information itself.
Use strategic imperfections to innovate design.
Strategic imperfections are imperfections that are employed intentionally to serve a higher purpose. We regularly make use of strategic imperfections in our website to test out design innovations. Think of these as the design equivalent of mutations in biology. Or to use a physics analogy, we operate our website at a low, but non-zero, temperature. In this way, we are able to optimize and evolve our site continuously, without the need for major untested revisions.
Re-learn print design to do better digital design.
Many of the principles for digital design were motivated by the coarse pixelation of early screens. These include the ubiquity of sans serif fonts, ragged right text, and the absence of kerning and word breaks. As pixels continue to get smaller and become sub-visual, the need for such ad hoc principles diminishes. Print design is a sophisticated craft that has developed over 500 years. By re-learning many of the tricks of the print trade, digital design at the London Institute benefits from increased legibility, consistency and beauty.
Create a science of funding science.
Given the importance of research funding, it is surprising how unsystematic scientists can be in how they approach it. In contrast to this, the London Institute has developed a science of funding science. We determine the full range of funders, play their odds of success, and modularize grant applications so that we can quickly recast them for different funders. At the same time, we recognize that a fundamental problem at the heart of competitive grant schemes is that the opportunity cost of writing so many failures can be comparable to the value of the grant itself. For this reason, in parallel to grant competitions, we also seek business sponsorship for basic research, described below.
Know the odds of success for research grants.
The success rates of research grant competitions vary from 2% to 40%. Yet many researchers are oblivious to these success rates. It is essential to know the odds of winning a grant, so that you can maximise the return on your time and know when applying is not worth the effort. We collect the odds of success for national and international funding competitions here. If an agency or foundation is reluctant to disclose its odds, it is typically because the chance of success is too low to make applying worthwhile.
Know the full range of research funders.
Most scientists are aware of only a fraction of the agencies and foundations that fund research in the mathematical sciences. Beyond the EPSRC, both the BBSRC and the MRC are sympathetic to quantitative investigations. But there are many other funders outside of the Research Councils. These include the myriad of programmes under Horizon 2020, the Royal Society, the Leverhulme Foundation, the European Research Council, the Wellcome Trust, the Simons Foundation and Cancer Research UK. That’s not to mention US and UK defence funders, described below. If you have taken the time to write a good research proposal, you should submit it in parallel to as many funders as possible.
A lot of basic science is funded by defence.
The London Institute has a strong track record of working with the US Department of Defense and the Ministry of Defence. This accounts for a third of our funding, all of which has been for basic science. Our agility and defence experience means that we can negotiate contracts and deliver results fast. The Institute’s first investor was DARPA, and other defence agencies that fund basic science include the nascent ARPA UK, the Defence Science and Technology Laboratory, the Office of Naval Research, the Air Force Office of Scientific Research, the Army Research Office and the Defence Threat Reduction Agency.
Businesses support basic science too.
Counterintuitively, sometimes businesses are more open to sponsoring curiosity-driven research than government funding agencies. Government agencies often over-emphasize predictable, short-term impacts. Businesses, by contrast, are sometimes willing to invest in the development of new, more abstract ways of thinking that they would be unlikely to realize in-house. Some businesses don’t expect the research to benefit them directly at all. Instead, they fund it as part of their commitment to giving for the public good. We have a good track record of corporate-sponsored, open-ended research, particularly from the Boston Consulting Group, Bit Bio and the Centre for Defence Enterprise.
Take indirect costs seriously.
Research grants are broken into direct costs and indirect costs. Direct costs cover the direct scientific activity of the research project, such as the scientists’ salary and equipment. Indirect costs cover the indirect needs of the project, such as rent and science communication. Because the London Institute doesn’t receive student fees or government subsidies, covering our indirect costs is essential to our survival. We negotiated our indirect costs rate with Research Councils UK, and we use this rate for grants from other funders. Some agencies, however, cover no or limited indirect costs, a fact that they advertize as though it were a virtue. But this is far from admirable. It’s like offering to cover the cost of an operation, but only paying the surgeon’s fee and dismissing the cost of the hospital. In the same way that surgery requires more than surgeons, research requires more than researchers.
Papers and publishing
Don’t publish papers you are not proud of.
A scientist’s papers are like an artist’s paintings. They are how the world knows you, today and for posterity. Resist the temptation to publish a paper merely for career advancement. Papers that are insubstantial, in a faddish field or poorly written are bad for you, bad for the London Institute and bad for science. Instead, publish papers that are inspirational, clear and, if not always important, at least cleverly amusing.
Get feedback to avoid familiarity blindness.
It is difficult for writers to judge the clarity of their own work. When they read it through, they remember their intended meaning rather than seeing the actual meaning of the words and equations they have set down. At the same time, anyone with specialist knowledge of a subject tends to underestimate the ignorance of others. The best corrective for this universal syndrome is third-party feedback, and the best time to get it is before submitting a paper or grant proposal to a journal or funder. The London Institute encourages its members to read and give feedback on each other’s work, and to seek mock reviews from external peers, described below.
Pay to get papers and grants mock reviewed.
Getting critical feedback on papers and grant proposals is more easily said than done. First, criticizing someone else’s work risks antagonizing them, so superficial praise, while less helpful, can be more expedient. Second, it takes more time to criticize because doing so requires a detailed understanding of the work. The result is that it’s easy to find praise, but you have to pay to receive criticism. The London Institute pays for its papers and grant proposals to be mock reviewed before submission, offering £250 to a peer chosen by the author. This helps tame the noisiest aspect of research, peer review, by offering frank criticism before a paper or proposal is on the line for acceptance or funding.
Make papers as concise as possible.
Generally speaking, the amount of time that a person spends reading a paper is independent of its length. Either he reads all of a short paper, or part of a long paper. Writing a concise paper is harder, because it requires a more thorough understanding of the problem and its solution. Crucially, it requires the writer, rather than the reader, to determine what the most important insights are. Having to produce a tight narrative generally drives better science, and the end result is more likely to be taken in by the reader. It’s not for nothing that the most renowned journals have the most stringent length limits.
Promote papers in line with their value.
The cost of a research paper is typically between £100,000 and £250,000. This includes the cost of the researchers’ time and the indirect costs connected to the research. There is no point spending so much money on a discovery without making a concerted effort to get it in front of those who are most likely to build on and use it. The London Institute invests in promoting its papers in line with their value. If they are newsworthy, our in-house writers translate them into accessible accounts for the press. Some journals request a fee to make a paper open-access and free to download. Typically this fee is a couple of percent of the cost of the paper, which the Institute pays as a matter of course.
Avoid unchecked open-access publishers.
A brand is a promise of repeatability. Journal brands signify that papers published by them tend to meet certain standards of importance, clarity and broad interest. A brand is a decision-making shortcut, which helps readers decide which papers to read when they don’t have time to assess them all. We aim to produce papers that are important and clear and often of broad interest. For this reason we oppose the use of unchecked open-access journals, such as PLOS One and Scientific Reports, which publish on the basis of correctness alone, without regard to a paper’s scientific value. Very occasionally it may be necessary to publish in such journals, for instance, when a paper presents a heretical result, which wouldn’t get a fair hearing elsewhere.
Make seminars more fun and effective.
There are too many science seminars and they are often dull. The result is that the typical turnout at a seminar is low. This is bad for morale and a waste of time for the speaker, who will have spent over a day preparing, travelling and talking. At the London Institute, we try to make seminars more fun by turning every seminar into a party. We start at 5 o’clock and, crucially, serve drinks before the talk as well as after. This makes for better questions, since self-consciousness is soluble in alcohol. Informal discussion continues into the evening, which is where the real opportunity for serendipity occurs. We also try to kill two birds with one stone by inviting speakers that we would in principle hire. This keeps us abreast of the talent pool and is a great way to get to know candidates.
Regularly revisit how we communicate.
Because we trade in ideas, the way in which we communicate them is especially important to our impact and perception. Every Friday we meet at 5 o’clock to assess and rethink how we tell the story of the Institute and its science. All aspects of communication are on the table, from our website and papers to our press coverage and role in shaping science policy. This meeting is especially for those involved in our communications, but all members of the Institute are welcome to attend.
Get everyone together each week for drinks.
Every Friday we have Institute drinks at 6 o‘clock. This follows on from our communications meeting described above. Drinks take place in our seminar room, which converts into a fully-stocked bar. Everyone at the Institute is invited, and visitors from outside the Institute are also welcome.