Talking about how we talk about G-Research
When Mel Conway first penned his now-famous paper, he figured he was on to something, and sent it off to the Harvard Business Review for publication. After all, it was a breakthrough insight into how organisations – well – organise themselves.
But HBR refused to publish it. The serious periodical claimed that Conway had not proven his thesis: that the way institutions behave follows from their form. This fundamental truth of modern management, codified reality in a concise and novel way, was just too simple to be convincing. It strained credulity:
Any organisation that designs a system (defined broadly) will produce a design whose structure is a copy of the organisation’s communication structure.
That simple insight, so eloquently expressed, became known as Conway’s Law, and would help management-thinkers process an array of institutions. Nowadays, most tech-savvy people have a passing familiarity with Melvin E. Conway’s seminal paper in the 1968 edition of Datamation, the same paper that the HBR rejected.
Our organisation, G-Research, is striving towards greater transparency in how we do what we do, not just so that we can keep attracting the best minds from around the UK and the rest of the world, but also so that we can begin talking about our technical achievements to better engage with the communities that create the innovative software and architectures we need.
To build a better system we need to consciously change our communication patterns. This blog is my first step into that.
The successful company’s dilemma
At G-Research, we’ve had a history of pursuing innovations when it would help us solve a problem in the here and now, be it for faster processing of information, or the more accurate prediction of market movements. The demand for change was always responsive. We’ve consistently risen to the challenge too. It’s what has allowed us to stick around through waves of technological change and market turmoil.
But we’ve been hit by the successful company’s dilemma – namely that the things that worked to make us successful in the past, are not the things that will keep us successful, let alone lay the foundations for future successes. We’ve got to build bridges toward our future prosperity and ensure that we have the technology in place to take us there.
Companies that rely on software really feel the pinch of this problem. It’s so common, in fact, that we all have a name for it: ‘technical debt.’ Once you build something that works, everyone is afraid to mess with it, for fear of sending it off-kilter. Or they are busy adding the next new feature. What was once an asset can become a liability.
Martin Fowler discusses the differences between types of technical debt in his book about software design, ‘Refactoring,’ in which he describes the ‘Technical Debt Quadrant.’ His point is that design-debt is inevitable, as technology advances. And it’s a fair one. When we started building the G-Research computation platform back in 2001, many finance companies ran Microsoft Windows and were interested in C# as an innovative, new programming language, so that’s what we chose. If we’d known then what we know now – namely that there’s so much more innovation and development happening in and around open source technologies – we would have chosen differently.
Today, we’re investing heavily and re-platforming to Linux so we can take advantage of the ecosystem around open source, data science and distributed platforms at scale. It’s a big change for us. Where once the very fact that we used quantitative techniques to make predictions was something very few companies did, we now work in a world where data and predictive analytics are everywhere. That means we need to change our approach and look more to the outside world for inspiration.
But some things stay the same. Our core values have remained constant and will continue to guide what we do.
We believe in the power of data to provide insight. You’d think that goes without saying, given our business, but many people don’t appreciate just how important data is in making decisions. There is a pervading myth that financial markets are driven by some sort of vague “animal spirits,” or a kind of magic involving technical analysis-charting. We know better.
We believe in what the economist W. Edwards Deming said: “In God we trust; all others bring data.” It’s a clever quip, but it doesn’t stop with having the data, it’s also about what you do with it. And we’ve built some remarkable predictive models that have performed extremely well for our clients. Those models rely on smart mathematics – very smart mathematics – and a rigorous, intellectually honest process. Recently, we’ve spent more time and effort on using a data-led approach to measure features of our own company – and have made great improvements to processes, systems and how teams are organised simply by using what we’ve always known about data and about ourselves.
Playfulness fosters creativity
We work hard to make G-Research a fun place. It may seem counter-intuitive, that more fun can produce more work, but I look at it like an investment. One that pays dividends.
We’ve found that our team members enjoy a wide range of pastimes. We love that they are able to take part in games such as Mahjong, football matches, squash ladders, lunchtime runs, and charity events (such as racing up all the stairs in the Gherkin – madness). When asked by new joiners, or in interviews, about what sums up working at G-Research, I’ve usually said ‘the fact that we work hard on challenging problems, but have fun while doing it.’ It is really important to me that we continue to offer this kind of experience to technologists who come to work here. It also avoids burnout.
Encouraging people to enjoy themselves at work and with colleagues outside of the office helps our team to work creatively together to solve the kind of fascinating complex problems we tackle every day.
Focus on what matters, invite collaboration
As CTO, my focus is on connecting the dots between what is happening in the outside world and how that can be applied to what G-Research does. We have a long history of solving hard problems and providing platforms and research to our clients that allows them to be successful. The world is changing – and we need to move fast to stay out in front.
This means we need to constantly look at what hardware, software tools, development approaches or operational practices are available. Who is succeeding? What is becoming old hat? Where will our next pile of unexpected technical debt come from? What fundamental limitations can we find ways around? Now that data science has been democratised, we need to be a lot more engaged with the outside community.
So I’m going to begin communicating more about the technical decisions that we make, the reasons behind our choices, and how things got to be the way they are as best as I can. It’s a step toward transparency and openness – to begin believing that sometimes the smartest people in the room are the ones waiting for an invitation to the table, looking for an opportunity to contribute. I’m looking forward to sharing thoughts and hearing your insights.
Chris Goddard, CTO