*Update: I’ve added a lot of new information based on feedback.
This essay is an attempt to share many of the insights the team at Web Summit have learned, in the hope they can be useful to those interested in bringing people together in small or in very large numbers, for family gatherings or for company gatherings, for industry gatherings or for community gatherings.
What follows is an explanation of how a group outsiders, with no background in conferences, used a combination of software and hardware, a sprinkling of maths, and a dash of design, to create in six short years the “largest and most important technology conference on earth”, according to a journalist we got drunk on a pub crawl.
More seriously, six years ago Web Summit started as a tiny, local tech conference with 400 attendees. We had no idea what we were doing at that first conference. But six years later, more than 50,000 people from over 160 countries travelled to Web Summit in Lisbon.
For some, like USA Today, Web Summit has become the “largest tech conference in Europe”, for others like the Guardian it’s become “Glastonbury for Geeks”, and for others still like Bloomberg it’s become “Davos for geeks”.
Puffery aside, whatever Web Summit has become, it’s perhaps “the fastest growing conference in any industry ever” and the reasons why should be of interest to anyone who organises events, be they conferences, off-sites or even pub crawls. So let’s begin.
Humans are a deeply social animal. We’re constantly gathering together to eat, to celebrate, to play and to network with each other in the offline world. Together amongst friends we thrive; alone, cut off and isolated from human contact we wither.
In the online world, Facebook, WhatsApp and similar platforms have connected us to together in previously unimaginable ways. But the magic of Facebook is powered by a seldom-written-about field of mathematics called graph theory. Without graph theory those sometimes freaky friend suggestions on Facebook and their incredibly addictive News Feed simply wouldn’t exist.
A graph is merely a network, like a phone network or social network. And like most networks, the utility of Facebook is proportional to the square of the number of it’s connected users. In other words, as the number of Facebook users grow, it’s utility to users should grow exponentially. If none of your friends were on Facebook, then what would be the point of using it?
Graph theory makes Facebook powerful. Conferences, in my mind, should be no different, and yet they are.
Graph theory has never really been applied to improving how we connect in the real world at conferences. Despite a decade of social networks, conferences have not really changed all that much. That’s particularly unfortunate because there’s so much that graph theory can improve.
Ultimately, as the number of people from your industry at a conference grows, so too does the probability that there are people worth meeting. The problem however is finding them, ideally without even having to search or filter for them.
Web Summit has started to solve that problem.
Without going into too much detail at this point, instead of hiring experienced event managers, we’ve focused on hiring academic physicists with backgrounds in complex systems. Instead of manually curating who should speak at Web Summit or even where you should sit at a dinner, we do it algorithmically.
For example, out of the more than 1,000 combinations of talks and workshops at Web Summit 2016, our recommender systems cut through the noise to find what’s likely of most interest to you. On the left you’ll find my recommendations for the first day of this year’s Web Summit. Those recommendations are based on an early version of our schedule, which is only 50 percent complete, and using a test version of our iOS app and updated recommender system.
You can immediately see, if you know me that is, that it’s finding talks right across our more than 20 stages and dozens of workshops that really are of interest to me. I don’t even need to browse or search. Our recommender systems figure out what I might like by crunching billions of data points, but more on that later.
When we first gathered 400 people together in 2010 for Web Summit, using graph theory just seemed logical. There’s actually nothing complex or novel about our approach, but as an approach it seems to work for most people who attend our conferences and it should therefore work for all conferences. Back in 2010, word spread after that first tiny gathering in Dublin that Web Summit was “kinda okay”, and and in time that word of mouth has catapulted Web Summit to become, in the words of another journalist who enjoyed their pub crawl at Web Summit, the “largest and most important tech conference on earth” in six short years.
Independent survey data shows that “personal recommendations” drive new attendees at Web Summit more than the next biggest reason by a factor of 12.3. More than 80 percent of people purchase tickets based on a personal recommendation, compared to just 6.5 percent based on advertising, followed by a long tail of other reasons.
Any organiser of a gathering large or small, can do so much to improve attendee experience by applying a little math. What follows is a more detailed explanation of what we do, providing insights into not just graph theory, but it’s application in our use of software and hardware. I’ve also shared a further insight into how we work, through the window of lanyard design.
I hope overall this essay helps anyone interested in bringing people together, be that friends and family, employees and colleague, communities and industries. If you’ve any questions just email me at email@example.com.
An understanding of relevant approaches to mathematically modelling human gatherings, which are networks, is an absolute prerequisite of improving conference experience for attendees, especially as a conference scales. Graph theory, a field of mathematics first developed by German mathematician Euler, is the fundamental building block for optimising outcomes for attendees at conferences.
But improving conferences also requires attending to the many details that combine to create this moving, breathing gathering of human beings we generally refer to as a conference. Lanyard design is, I believe, instructive in that regard.
If you aren’t aware of at least some of the basic principles or science related to the design of a lanyard, I think it’s challenging to create a good conference or gathering of people. When it comes to lanyard design, an understanding in particular of typography, anthropometry and way-finding are important when establishing ratios, hierarchies and more for your lanyard, but more on that later.
In my view, if a conference organiser is not regularly reading academic papers on topics related to network science in particular, as well as crowd psychology and dynamics, way-finding, anthropometry and related areas; running controlled experiments at their conferences and conducting rigorous surveys of attendees of all types, I believe it’s challenging to improve each iteration of your conference.
As a maxim of conference design, the smallest of details thoughtfully attended to generally combine to create a better conference. A poorly designed lanyard, which normally holds your name badge, is more often than not indicative of the overall level of rigour applied in the creation of that gathering and consequently a reasonable predictor of likely overall conference quality.
Zooming out for a moment from a component of a conference, like a lanyard, to an entire conference, I’d suggest that a failure in particular to appreciate that human gatherings are in themselves complex systems, will limit the ability of an organiser to improve overall conference experience or outcomes for attendees. In particular as a conference scales.
To better understand why graph theory is important to conferences, it’s useful to consider its core utility in helping optimise the product offerings of two of the most disruptive companies of the internet age, Facebook and Google. Facebook’s ability to suggest likely friends and engaging content and Google’s ability to surface relevant search results, through PageRank, rely on graph theory
Similarly, Web Summit’s ability to surface heretofore seldom heard speakers across a multitude of fields, recommend relevant talks to you, place you in relevant pub crawl groups, meetings and roundtables, as well as suggest to you people to meet, both through our apps and in person through our staff, also hinges on graph theory.
Below you’ll find a very outdated visualisation of how we identify communities and surface speakers. This visualisation is over two years old, and was made to help some of our internal teams understand how some very simple maths can go a long way in finding, for example, relevant speakers from the Bitcoin community. The process we followed in 2014 for identifying a seed group, finding associated nodes, and thereafter computing some basic network statistics was incredibly simple. The volume of data we used for this visualisation is also incredibly small, though I hope it’s nevertheless useful. In the final segments of the video we use a Force Atlas algorithm to lay out the graph and cluster sub groups, and then a Fruchterman-Reingold algorithm to bring the most influential nodes to the centre of the graph. If this sounds complicated, it honestly is not. A cult of obfuscation has built up in startup-land, where the go-to words have become algorithm, machine learning and artificial intelligence. It’s just maths.
Another useful example of graph theory in action at Web Summit are our dinners. Let’s say there are 150 people due at a dinner you are attending at Web Summit. To help optimise which table you might be assigned to, we rely not on the information of the 149 others due at the dinner, but instead on our own graph, which we refer to as our meta-graph, containing approximately 130 million people at present and billions upon billions of other objects derived from public and private data sources.
At a macro-scale, this data allows us to peer deeper into who you really are and how you fit into the world of Web Summit, in particular what dinners we should invite you to and to which tables we should assign you.
At a micro-scale, we also consider all the interactions you’ve had so far related to or at Web Summit, past or present. This can include, but is not limited to, who you’ve chatted to, whose profiles you’ve viewed, which talks you almost certainly attended or planned to attend, who messaged you, who viewed your profile, location information, the hotel you are staying at, your activity on all of our emails ever, and almost every other data point you create through Web Summit or leave a trace of on the public web.
Assigning you to a table at a dinner is one challenge that graph theory can help optimise. We go further at times at Web Summit by assigning you to a particular seat using a little simulated annealing. This system works particularly well for small groups who attend a number of overlapping dinners, lunches and other small events, where there’s a potential for a number of no-shows at each small gathering.
Turning to Facebook for a moment, the manner in which Facebook’s engineers approach the challenge of surfacing relevant friend suggestions or content, while predicated on graph theory, is somewhat different to Web Summit.
While most of our recommender systems for attendees at Web Summit rely on mathematical approaches similar to those used by Facebook, the recommendations we are seeking to surface for attendees are often the inverse of those of Facebook. Facebook mostly attempts to mirror your likely already existing real world network, suggesting to you friends you likely already know. Web Summit on the other hand more often attempts to find potential structural holes in your real world network. From there we work to engineer serendipitously a potential conversation, meeting or otherwise with another attendee.
A Web Summit Pub Crawl
You’ll feel some of our deliberate nudging when using our app, receiving our emails or seeing our advertising, but other nudges are simply unseen and manifest themselves in different ways over the course of your attendance at Web Summit. In short, in-app recommendations and dinners are only two of a wide array of use cases.
Internally, we’ve built a suite of software for sales and marketing predicated on our meta-graph and further applying a little maths from other relevant areas. As an example, when it comes to ad-tech, we are the largest user in terms of throughput of Facebook’s advertising API in Europe. Those ads you may forever see, often represent only one of 10,000 variants created in a single week, and each of those variants may be shown only to a tiny subset of the total users on Facebook.
While gaming companies and travel companies, for example, rely to a large degree on humans to create ads, we borrow from areas like combinatorial optimisation. Then when it comes to monitoring, we have our own reporting stack, which monitors performance in real-time, auto-scaling combinations of ads that are performant and killing those that are not.
What we’ve built is by no means perfect, but it is almost certainly more sophisticated than any other available ad-targeting technology for the needs we have. The targeting sophistication is greater not because we are any more or less brilliant. But instead because the targeting sophistication of ad technology platforms is, by design, blunted by their need to appeal to a wide number of potential users or advertisers. In our case, we can fine tune our own ad-tech stack to do exactly what we need it to. Ad-tech companies build a sort of one-size-fits-all golf club that has mass appeal, but lacks precision. In our case we customise our putter to the nth degree for our own highly specific needs.
As a side note: I’d use the words algorithms, machine learning and artificial intelligence more liberally as is de rigueur, but from my basic understanding they seem somewhat overused, misunderstood and overstated these days.
Before specifically focusing on lanyards, let me finish by outlining some other use cases of our general approach as I believe they will be helpful to anyone involved in or considering gathering people together.
An enterprise software company that exhibits at Web Summit may be looking for leads similar to their existing customer base. If that enterprise software company is prepared to provide us with a seed list of existing customers, combined with as much relevant data such as company size, contract value and more, we can identify say the best 500 leads from say a total of 50,000 attendees. We can assign a weighted probability against each lead and rank them accordingly. That company can then prioritise outreach to these leads ahead of or during Web Summit. We can do this for leads at a company level or at an individual level.
Some enterprise software companies coming to Web Summit 2016
While this approach spooks some companies, it has broad application. In 2014, amazed at the potential of advertising on Facebook but frustrated by their lack of targeting sophistication at that time, we built a similar tool. As people booked tickets and connected with Facebook, we used a variant of eigenvector centrality to estimate the most likely subset of a person’s friends to next book a ticket to Web Summit. We then created highly personalised ads showing you pictures of friends who had already booked tickets. If you clicked on this highly personalised advert, you were brought to an attendee page that dynamically generated an optimal attendee list based on the specific ad you had clicked combined with any information associated with your IP address.
The above approaches are in no way proprietary. Every element uses only known mathematical approaches, and can be replicated easily. The only constraint is time. However, if you want to skip building your own graph and a layer of applications on top of that graph, you can always try tools like EverString, who have raised nearly $100 million to productise the creation of company lookalike audiences. In more recent years, Facebook have productised lookalike audience creation for advertising and vastly improved the sophistication of their targeting. Finally, if you want to try and personalise your website and don’t speak API or related jargon, Optimizely have an out-of-the-box solution.
Lastly, I just want to detail a final use case of our overall approach, as I believe it will help other conference organisers, particular organisers of technology conferences.
At Web Summit, for investors wishing to meet relevant startups and startups wishing to meet relevant investors, we’ve learned over time that we need to look for somewhat different types of signals amidst an ocean of noise than what might first appear obvious.
Chris Sacca speaking at our new conference Collision in New Orleans
For example, we now begin to infer startups with a high likelihood of interest amongst certain or all investors before Web Summit based on a combination of signals including, but not limited to, all available public data on these startups and investors, as well as advance meeting requests, messaging activity, profile views and more through our app. In real time, as data is added to our meta-graph we re-weight our recommendations to investors of startups to meet and to startups of investors to meet.
We supplement in-app recommendations, with a team of human beings before and during Web Summit who help guide investors and startups to their own self-interest.
To any investor reading this, our graph, where possible, contains data on every investment you have ever made that we can reasonably track, the likely competitors to all of the companies you have backed, individuals and funds with whom you have co-invested, those you share board seats with, everyone you follow or are friends with across a wide range of social networks, the chronology and reciprocity of those connections where possible, the type of content you share online bucketed using some NLP, and an array of other useful data points that help us orientate you to your own self interest. We furthermore observe all of your initial meeting requests and compare these to those made by every other investor, attendee, journalist and more. By combining billions of data points we infer probabilistically some somewhat useful insights. These insights allow us to help make your job easier and ultimately make better use of your time and the time of startups.
Below you’ll find a screenshot of one of our internal tools The Lookalike Machine. In this case, VentureScore is a measure of demand within our community to hear an investor speak. This relies on a large volume of data, more on which I will share another day.
This is a screen grab of one of our internal tools for ranking investors by a given score
From experience, our investor recommendation system results in rapidly expediting the natural herding or flocking behaviours of Silicon Valley’s leading investors. At Web Summit the best startups will end up with exponentially more meetings with the best investors. We work to proliferate awareness that a given startup appears interesting to the most relevant and/or best investors. We have our own methodology for scoring the “best” or most relevant investors, which we will share another day.
We may even have found a nice little power law in venture investing
However, while this may sound sophisticated, it’s really not. Furthermore, a note of caution: while our data-powered investor-startup and startup-investor recommendation system has two upsides, it also has one downside.
The first upside is that some startups benefit significantly more than others from our overall methodology. Marketplaces are by their nature meritocratic, and not democratic. The better the signalling for a given product, the greater the interest. Startups that gain early traction at Web Summit can oftentimes end up with a disproportionate number of meetings with investors, sometimes in excess of 100. This experience will likely spill over into meetings with journalists and other attendee types.
The second upside is investors with a strong track record benefit from both the wisdom of the crowd, and billions of data points that we have gathered and organised. We help investors cut through noise to find the signal, matching them with more relevant startups or with simply “hot” startups before they may even realise a startup is hot. We do this in a very frictionless, non-obvious way. That said in any given year the number of startups an investor might look at and not invest in, must be greater than 99 percent. In other words, less than 1 percent of startups an investor looks at, or perhaps even meets, they subsequently invest in. This then is the base efficiency rate we are working from, which means even though our approach might be an improvement on absolute randomness, in absolute terms it may not result in a huge improvement in outcomes.
Finally, there is a downside for some startups and some investors. For some startups at Web Summit 2016 the number of interactions they will have with investors will be more than zero, but not much greater than low single digits. The reason for limited traction, to reiterate, will be because all the signals we can track indicate low probability of success in meeting investors. This does not mean these startups will not succeed in and of themselves as businesses, but it does mean that for whatever reason, investors are exhibiting low levels of interest and startups are displaying patterns indicative of a low-interest startup.
A further note of caution: we only know what we know. Our recommender system is based, no matter how real time, on historic data, upon which we make predictions. Every year great engineers create entirely new products. Those products often don’t fit a category or a pattern. They are non-obvious. Our system therefore is not infallible. It’s merely better than the chaos and randomness of a large conference unmediated by graph data and recommender systems. So yes we can make, with varying degrees of confidence, useful predictions, but no we can’t predict everything with absolute confidence.
A small tip to startups: if you message dozens of investors and no one responds, we start to deprecate you in terms of how often you are recommended to investors and other attendee types, like journalists. In other words, the chances of your startup being recommended to an investor start to fall. We’ve also bucketed you into one of about 20 industries. If you are messaging investors focused on industries or geographies unrelated to your industry or company location, then that indiscriminate messaging of investors will result in a highly discriminatory de-ranking of your startup. Of course, if you actually do your research and send highly engaging messages to relevant investors, then this signal will likely dramatically increase the number of times your startup is shown to investors. The same applies for every attendee type.
For some investors, if you make zero meeting requests in advance of Web Summit and we furthermore cannot infer much about you as an investor, the quality of startups we will recommend you will be lower than those we push towards say an Index Capital or NEA. Investors with great track records can all but show up having done no groundwork and they may still get assigned meetings with “hot” startups. Again this may appear as unfair, but ultimately marketplaces are not democratic, they are meritocratic.
In a very real sense, we re-assign recommendations of and meetings with startups with limited traction to investors with a limited track record. This is a highly efficient market mechanism as compared to the absolute chaos and randomness of normal conferences.
Every month we get better at surfacing recommendations. Where we are today as compared to Web Summit 2015 is incomparable, but still a long way from anything approximating perfect. Below is an example of another internal tool that we call Rankatron 3000. We use it to rank communities, like Data Science. In this case, we identify the most important data scientists in the world in the eyes of the data science community.
These are the most important data scientists in the eyes of the data science community
Another cautionary note: while the volume of capital that has been invested into the many startups that have attended, exhibited or pitched at Web Summit appears almost perverse, correlation is not causation. Many of the startups in attendance were almost certainly on a path to success. The startups exhibiting at Web Summit are a deeply unrepresentative sample of startups in general. To understand just how unrepresentative startups at Web Summit are of startups in general, this week we passed 450,000 tracked startup applications over Web Summit’s history. Every startup goes through a review process, where they are scored directly and indirectly. Those that end up attending skew heavily towards those that score highly. In four years, only 1,070 of those 450,155 startups have ever scored more than 90 percent through our assessment mechanism.
While we don’t always appraise startups correctly during our assessment, and while we sometimes get it very wrong, a reasonable portion of the startups we select have a higher-than-normal probability of future success.
Finally our historic scored dataset, combined with other data points, helps us sharpen to a point our ability to appraise each new startup that applies. But again our approach is far from perfect.
At this point, I just want to say that I have no idea how people make use of large conferences, unaided by graph theory. I also fail to understand why for so long human gatherings in the real world have remained static, while human gatherings online have benefitted from all the wonders that a little maths combined with a little technology has to offer.
Web Summit aside, most real-world human gatherings have remained largely static and antiquated in their underlying approaches for millennia. Software and hardware, with a little basic maths sprinkled on top, should, we believe, change that. We’re attempting to test that hypothesis and we believe our general approach has merit, especially as conferences scale into the tens of thousands of attendees.
Here’s another way of thinking about it: the utility of Facebook is proportional to the sum of its users. As conferences grow their utility should similarly grow exponentially.
The larger the number of people from your industry in one room, the higher the probability there are people worth meeting. The challenge, however, is that unaided by technology, as a room fills with more people from your industry the probability of finding the people most relevant to you plummets. Intuitively sensing that is what I think drives most people away from large conferences. People are correctly associating large conferences with chaos and inefficiency. And that’s due to the failure, to reiterate once more, of conference organisers to simply use some very basic maths, combined with software and hardware.
Now that you have some basic sense for what we do, it’s time to go a little deeper on lanyard design.
Independent surveys show that about 2 percent of attendees at our conferences do not intend to return. I’d suggest that this is not just an unusually high level of satisfaction for the product category known as conferences, but that this level of satisfaction is the driver of our unusual growth. And behind that driver lies a general methodology, a methodology that can be even better understood by considering how we approach lanyard design.
With that in mind, I’m going to share some basic principles of better lanyard design. In later posts I hope to share some principles related to better queuing systems and a largely forgotten Danish mathematician, internal conference signage and a seldom celebrated Dutch designer who taught at RISD, how specifically to implement recommender systems for better table and seating assignments and other useful building blocks for better bringing people together.
Most companies create very singular products. In our case, we create many products, from better lanyards, to better registration software, to better recommender systems, that all combine to improve a product category known as conferences.
Explained another way, we’ve built lots of singular pieces of software, as opposed to lots of lines of code adding up to a singular piece of software. Our registration software focuses on speeding up registration for attendees. But our registration software doesn’t need to speak to our GoPros mounted in the ceilings of our conferences. Their role is to use a little computer vision to attempt to improve overall conference layout and crowd flow. As such, our engineering teams can work in isolation from each other. In many cases, they don’t need to be teams, just individuals curious about solving real world problems using software and/or hardware.
The only thing that might unify all of our combined data at a given point, is our meta-graph and various layers of API’s that manifest themselves across our attendee facing mobile and web apps and websites, and our internal software used by our teams.
To design something for human use, it’s important to reduce the object you are creating to its basic functions or principles. In Kantian philosophy these might be referred to as postulates, or in maths as first principles.
If somewhat complex equations, products or arguments are not predicated on defensible first principles, then they are often baseless. Most lanyards are baseless. They are chosen arbitrarily without any fundamental consideration for basic principles.
Those principles should in my view be grounded, first and foremost, in evidence-based research and only thereafter be modified by testing. The inverse process is incredibly inefficient, in particular if credible research already exists. It’s akin to reinventing the wheel, when people have already expended huge amounts of time creating, researching and refining that object.
There are three areas of research that stand out, in my view, amongst a long list that when applied in combination can at least inform lanyard design: way-finding, anthropometry and typography. Let me first explain each, and then discuss how they inform lanyard design.
What is way-finding?
Way-finding “refers to information systems that guide people through a physical environment and enhance their understanding and experience of the space. Way-finding is particularly important in complex built environments such as urban centres, healthcare and educational campuses, and transportation facilities”.
Conferences are complex environments. Lanyards, which hold name badges, can help.
What is anthropometry?
Lanyards should be designed not just with the shape of the wearer in mind, but with the field of vision of the reader in mind too.
What is typography?
Typography “is the art and technique of arranging type to make written language legible, readable, and appealing when displayed”.
Lanyards are worn primarily to be read, typographical choices should therefore, in my view, be informed by research on legibility.
Lanyards should first and foremost increase human-to-human interaction at gatherings. They are, in a very real sense, a visible calling card, sign or business card, normally worn around the neck of an attendee.
Lanyards should therefore be optimised for ease of legibility by other attendees from an expected pre-conversational distance and should furthermore ensure interactions are as natural and human as possible.
To illustrate this point, when engineers place signs on highways in the United States, the typographic choices informing those signs apply principles that take into account the likely speed, distance and viewing angle of drivers. Other considerations include the impact on legibility of the number of letters per line, the line height-to-width ratio, kerning, line spacing, font weight, capitalisation and font to background colour contrast, amongst others.
In 2004, in the United States after extensive research the font was switched from Highway to Clearview. In recent months there’s been a reversion of sorts. In Ireland, where I live, we use a font called Motorway.
When it comes to airports, subway systems and bus stations, there’s thankfully a huge wealth of knowledge to draw upon, spread across a number of academic journals and books. This knowledge informs everything from the typographical choices and hierarchies for our lanyards to the contrast ratios and kerning.
In our typographical hierarchy, for example, the single most important word on your badge is your first name. We have deliberately shortened the length of lanyards from their near-standard length so that your first name sits, on average, within a more natural line of sight even when speaking with someone. In other words your name badge does not hang somewhere close to your belly button, as is the result of the near-default lanyard length used globally, but instead on your chest.
The font type, style, weight and size of your first name is optimised for legibility at distance. However your badge is also small enough so as not to overwhelm you. A huge influence on many of these choices is this amazing book.
There are a number of other considerations. The cheapest option for badging is a plastic pocket with lightweight paper inside of it. These tend to bend, crease and damage quite easily. Legibility is reduced by the sheen or gloss of the plastic. They are also difficult to store outside of the conference. And furthermore, the weight and overall experience is sub-optimal in our largely amateur view.
For all the above reasons we have chosen a seldom-used but durable composite material, which roughly equates to credit card plastic. Our badges are largely splash proof, tend to keep their shape and tend not to scuff. The exact size of your badge is also designed to fit in the median jeans back pocket, but not disintegrate when you sit on it.
Instead of sharpish corners, we now use curves or splines that help prevent your badge developing dog ears. While the cost of our lanyard design is probably about ten times that of a normal lanyard, it’s still cheap in absolute terms compared to the price of the ticket. Most importantly, we hope the return to an attendee of a more functional lanyard and name badge is worth the time and investment.
To reiterate, your name badge is a sign. That sign should trigger conversation with as little friction as possible. Ideally someone should be able to view your first name and start a conversation without visibly looking down. The exact location of your first name has been chosen based on the expected field of vision of another attendee. Perhaps in time we will create small, medium and large lanyards depending on the height of the wearer, but for now we’ve settled on a median length.
Based on experimentation, your surname is somewhat irrelevant, as is your title. In almost all instance, first name is critical, company is a distant second.
We have deliberately chosen to remove an attendee’s title, in particular to create a trigger for conversation. “Lorraine, what do you do at Google.” “I run all marketing and communications globally, Tim”. “And what do you do at Apple Tim?” You get the picture.
Our lanyards represent many years’ work, but we are still at the start of a longer journey. We have now run field experiments with four companies over the last 18 months, using a mixture of technologies including RFID, NFC and some frighteningly advanced chipsets in partnership with the amazing Bell Laboratories.
These technologies allow us to extract incredible amounts of data from your badge in real time, which we then append to our meta-graph to further improve our recommendations to you in real time. Using advanced sociometric badging technology at Web Summit 2015 we could measure not just your exact location, but using some neat technology that has only been used at Web Summit the direction of your voice, which helps us identify who you are talking to, the cadence of your voice, which helps us gain insight into the likely context of the conversation, gesticulation using gyroscopes and so much more.
Lanyards are just one of many components of a conference. Our overall approach is to improve every component of a conference, constantly, and in a way that can eventually be applied to human gatherings of all sizes, for all industries and in all locations. We are forever in search of better approaches, systems, designs and more across our 10,000+ conferences, 1,000+ conferences and our 100+ conferences. No component is ever perfect, and the challenges of making a 100+ person conference better, are at times very different to those of a 10,000+ person conference. Most importantly, whatever we have achieved to date, we feel we are very much just getting started.
**Update: Additional Help For Event Organisers**
For those organising conferences who simply don’t have the time or resources to think or do any of the things mentioned above, you should at the very least consider using a white labeled event app. I think they’re great for anyone operating on a shoestring. Eventbase, DoubleDutch and Pathable are absolutely brilliant for organisers who just want to get something into the hands of attendees. While critics will say it is impossible by design for these event apps to ever provide highly relevant recommendations, they can nevertheless at least make your attendee list searchable and filterable, and match you to friends.
Finally, how we approach conferences is informed by my own broad philosophical approach to work and life, which based upon requests following an AMA on Product Hunt I have shared below.
In the absence of anything resembling scientific evidence on how you should live your life on a day to day basis to ensure you are happy and content, I adhere to an entirely arbitrary and simple daily routine. There is no way of validating whether the patterns or behaviours I follow on a daily basis have any merit whatsoever. So please do not in any way whatsoever infer that I am providing advice. I am merely indulging myself.
Each morning I ordinarily get up at 7.05am. I start cooking porridge at 7.15am and by 7.25am I am reading. In the morning, I use Google Scholar Metrics as my starting point to find interesting papers for a period of 60 minutes. This seldom used tool helps to surface interesting papers recently published across every academic field. I focus only on papers potentially related to the work I do. Even with these restrictions the potential areas of reading are vast, and can range from network science to queuing theory to way-finding to office design to organisational behaviour and so on.
I don’t tend to read any paper outright, instead I focus on the abstract and conclusions/results. Sometimes I read parts of a paper related to the methodological approach of that piece of research. Particularly in the social and behavioural sciences the statistical integrity of research requires, at times, more careful reading. Having helped edit and establish some journals and academic awards across many fields, I get a little pernickety about these things.
Following morning reading, which I see as functional, I go to work. Broadly speaking I think many of the approaches to organisational structure, behaviour and communication which persist across technology companies are anachronistic in an age mediated by real time metrics on team and individual performance, as well as the many new tools for communication available to staff. However, my views on this are perhaps better for another day.
At night, I read from 10pm for 60 minutes. My focus is mostly on subjects somewhat unrelated to work: linguistics, number theory, economics, history, pedagogy, philosophy. The more obscure the better. This is an absolutely fascinating book on how we learn.
I avoid all books on best seller lists. If it’s in an airport book shop then I’m not reading it. I accidentally read a book on check lists a few years ago. It seemed to take a paragraph from an academic paper and then using some Gladwellian devices stretch that paragraph to two hundred pages. It’s an inefficient way to learn.
If I want to learn fast, I don’t want narrative or other literary devices, I just want facts and footnotes, and a pen or pencil to underline things and scribble in the margin.
Outside of reading habits, I live with five people: my expectant wife, an academic statistician, an entrepreneur, a designer, and an engineer. We keep chickens and a dog. Overall I think there’s compelling evidence that communal living is, when managed correctly, a very enjoyable and healthy way to live.
In general, I adhere badly to a form of material asceticism. I own one pair of jeans. All t-shirts that I wear have been given to me for free. My shoes included. My sweaters are made by my wife. I walk to work, but I do share a 15-year-old car with my wife. My only real expense in life is books. Books are my meditation. Underlining in particular, and leaving scribbles in the margins in general. If I want to travel in space or time, I read books. I’m not particularly numerate or literate, but just enjoy reading.
I see a purpose in doing things well and learning for no other reason than the joy of it. While I am not religious, I think my way of life was mostly influenced by a group of 50 monks with whom I lived alongside for 6 years of my life. They shared a passion for understanding the roots of all things, and above all living simply, unencumbered by material concerns.
I’d like to hope those monks approve of our lanyards.