As we head towards the prospect of a second wave here in the UK, Covid-19 shows no sign of going away.
At the same time, the social contract that has every citizen social distancing to protect the vulnerable is showing signs of fraying (e.g. Broadway Bradford lack of social distancing and masks concern), today's FT reported in Low UK isolation rates raise worries of virus spreading that self-quarantine and self-isolation compliance is as low as 11% and 18% respectively.
Governments are responding with traditional carrot and stick influence techniques (though mostly stick): fines for non-compliance, a reliance on rhetoric “we can do this together”, "a stitch in time saves nine" and a set of statistical graphs.
But with social distancing likely being required for the long term (or at least until a vaccine is actually distributed) are they missing a trick?
The science of influence has evolved considerably since those early carrots and sticks. Nowhere more so, than in the way in which we publish metrics.
"Gamification" is a design approach that offers a way that helps people both understand the numbers we publish and, importantly, act on them.
Can we use the principles of gamification to increase compliance with social distancing regulations? Can we provide better numerical feedback that encourages citizens to maintain social distancing for the long haul? Can the way government presents the data do better? In short, can social distancing be gamified?
In this post I want to show you how to apply the infinite gamification design approach to Covid-19 social distancing measures. With this guide you should be able to run a similar design session that will output a gamification program for your country, your region or your local area. With this toolkit you can publish numbers for the citizens whose health for whom you are responsible in a way that encourages them to take action.
Right now only fairly basic statistics (mortalities, cases, tests and the R number) are being provided (example above nationally and below locally). However these raw numbers simply aren’t that engaging for the general population. We tend to turn off. We become blind particular to the difference between 1,000, 5,000, 10000, and 100,000. When numbers get that big, one seems much like another. Too many numbers together and it just looks like a maths test.
Contrast this to the attention we pay when the numbers affect what we care about. How many football fans wouldn’t know the progress of their favourite football team in a league that is managed based on a number (points).
What if we could harness that same tribal passion to see “my local team” win at football and use it instead to drive good social distancing behaviour? This is the objective of a gamified approach.
To answer this question, I’ll be using the process outlined in my book, Infinite Gamification. I won’t be explaining each step in detail (that’s kinda what the book is for), so if you want to run the process yourself, then I recommend you have a copy of Infinite Gamification beside you as you do.
Before we start, let me be clear about the use of the word “gamification” - we are not making serious things fun, we are not turning real life into a game. Covid-19 is a deadly serious problem. Instead by “gamification” we mean using the underlying mechanics of games - such as points for example - to provide timely feedback that people actually take notice of. We provide the feedback in such a way as to encourage optimisation by the ‘player’ - the person or group the data relates to. It’s really thinking about numbers from the perspective of the person who is expected to change rather than from the perspective of the person who wants them to change.
A clumsy example of gamification is the classic “hospital league table” which posts a monthly statistics on how well hospitals are performing on a single metric. Hospital managers typically look at these league tables to make sure their hospital is performing well and will take steps to address this if they aren’t. For central government officials publishing league tables, this would seem a good technique to get hospital managers to focus on the areas that are important to central government. However, without good gamification design the approach can have unintended side effects - such as hospital managers turfing out sick patients too early in order to meet bed availability targets! [BMJ 2009;338:a3130]
Gamification is being used whenever you display personal analytics data to someone and expect them to change their behaviour based on that data. Whether you call it gamification or not, you are in effect “gamifying” them. The question then becomes whether you are doing it well - and driving good, healthy, optimising behaviours - or doing it badly - and suffering the consequences of unwanted side effects.
So with that in mind, we proceed to follow the Infinite Gamification design approach to ensure our gamification of Covid-19 social distancing is the very best kind of gamification!
There are three phases to go through toward a successful infinite gamification program - analysis (agreeing objectives and surveying the mental landscape for our players), design (identifying what feedback we will give) and evolution (how we will iterate and improve our program over time).
Analysis starts by stating what it is we are trying to accomplish and for whom - the prime directive.
The prime directive of this infinite gamification program is to encourage citizens living in the UK to maintain social distancing practices on a long term basis in order to curb the spread of the Covid-19 coronavirus.
The UK government makes data available online via a programmatic interface (API) at https://coronavirus.data.gov.uk/developers-guide - there’s plenty of data there to choose from. We will look at exactly which data points to use in the design phase.
So we can get the data but what about who we choose as the players - what are our options here? Are they regions (England, Wales), authorities (Bucks, Berks), cities, towns, post codes, streets, bubbles, families or individuals? Again there are more than one possible “player” type to choose from.
Like most data resources, the data isn't organised exactly as we would like it for our purposes, it's likely we will have to make some compromises. For example. the ArcGIS map shows data by postcode however the data API provides it by "LSOA" or "UTLA" - both of which seem to be something to do with local councils. Even the data itself isn't immune to peculiarities, for example no cases for a day is recorded as -99.
For this example workthrough I will be creating a gamification program using local councils (UTLA’s) as the players and use the coronavirus case data that the government provides daily for each UTLA.
Any live program implementation will require further analysis of the data. It's ok to stop the data sourcing process here though. One of the key principles of infinite gamification is that it is perfectly normal to iterate and evolve, we don’t expect to get everything in its ultimate format from the beginning.
National and local authorities are already providing considerable Covid-19 information - in this phase of analysis we look at what programs and scores are already impacting individual citizens.
Mapping some of the existing coronavirus and other local authority related individual scores gives us the context for any score we introduce:
The key point here is that there are already gamified scoring systems (some of them very binary - I have Covid or not) that are having a bearing on our players.
A key question to ask before we go any further is - are the existing scores sufficient? Do they address the Prime Directive sufficiently? Let’s remind ourselves of our prime directive:
The prime directive of this infinite gamification program is to encourage individuals living in the UK to maintain social distancing practices on a long term basis in order to curb the spread of the Covid-19 coronavirus.
Looking at the strengths and weaknesses of the existing gamification ‘programs’ we’ve identified helps us to see whether a new program is needed:
Meets requirements of the Prime Directive?
Have I got Covid-19?
A binary signal is not enough to drive the behaviour tweaking which a maintenance directive really requires. Without gentle nudges behaviour is likely to keep slipping, becoming more lax over time.
Is my area in lockdown?
A binary signal is not enough to drive the behaviour tweaking..
Did I remember my mask today?
Too hard to keep track of, data is not available. Smart masks that record usage automatically might help but are not yet widely available.
Too mathematical to drive individual behaviour - an R number is too abstract to encourage general behaviour change.
Case and death tally
Does not trigger timely behaviour change - by the time the indicators flag an issue it may already be too late to prevent a major outbreak.
Volume of Covid-19 media coverage
Does not provide a local signal - only a national one.
Extent to which friends and family are sticking to the rules
Not connected with objective data so prone to major errors.
It’s clear from the score context that while there are many programs in place which are guiding some aspect of behaviour, none of them are specifically targeting our prime directive.
However, while I have a relatively clear national picture I am unlikely to know my local situation unless I’m in lockdown. A good candidate for gamification would be to offer a local “canary” type program that might warn citizens in a local area that they are in danger of going into lockdown.
There is room in the market for a new, infinite gamification program, that does achieve our prime directive.
There are plenty of stakeholders to consider in our design - while the citizen is at the heart of the score feedback we want to give, we have to take into account local authority civil servants and NHS managers who will inevitably take a deep interest in their local score (if it is local authority based), headed up by local elected representatives. In the wider environment, there is media and national political interest - particularly in those areas that are not doing well and in danger of returning to lockdown.
As public health officials we are closest to the ‘Manager’ role in this project since we are keen to ensure community wide outcomes.
Since any of us are citizens and in effect, the players, we can do our needs analysis with a ‘self interview’ using the ‘focus group of one’. While ideally we would ask more people, we can get to our first program design just based on our own experience. We can then iterate later once others start to adopt the infinite gamification program for themselves.
We therefore ask ourselves the crucial question: “what feedback will help you achieve the prime directive?”
Here’s my own answer.
“As a citizen I want to get a sense of how well I am doing to preventing the spread of coronavirus. In particular I want to balance my desire to return to a ‘normal’ life, economic necessity and maintaining a safe environment for my community. I want to know if I can relax a bit and visit more public spaces or conversely if I am letting standards slip by close contact with too many people outside my household, not washing my hands enough or generally not social distancing enough. I want relevant, local and timely feedback so I can nudge my day to day behaviour in the appropriate direction.”
As a citizen Barbara is concerned to live her day to day life within the law, fulfil civic responsibilities such as paying tax and voting, and to notify authorities of broken services (potholes for example) and anti-social behaviour when it occurs.Ambitions
Barbara hopes for a quiet life enjoying the relationships of friends and family.Limits
Barbara shops for groceries locally and wants to enjoy face to face relationships at her local church and pub.
Barbara gets most of her news in passing from the BBC via Radio and TV. She doesn’t make a point of tuning in to specific news programs.Unmet needs
Barbara has little sense day to day of how prevalent the virus is in her local area and among her friends - should she be super cautious or can she relax a bit?Risk appetite
Barbara is personally fearful of catching the virus due to an ongoing health condition, she has very little inclination to ‘risk it’. If in doubt she stays indoors - Barbara has lived in the area most of her life and is not political. She tends only to engage with the local school and healthcare services.Demographics
Barbara is 40 and lives relatively comfortably with her husband and young family in a house they have a mortgage on.
The intrinsic motivation primarily is to stay healthy. Barbara does not want to catch the virus. A secondary motivation is that her local area and community should be successfully maintaining the social distancing regulations in order to protect the vulnerable around her. An extrinsic motivation might be ‘status’ - she would like to feel that her area is doing a good job of this, and keeping up a high standard in line with other local areas nearby.
Many gamification programs require extrinsic incentives - “win a new set of goal posts for your school” for example. My view is that these should be added on afterwards on an ad-hoc or ‘surprise and delight’ basis. As soon as you structure the motivation of your program around a specific incentive you immediately tie yourself in to offering that incentive long term. Far better to structure the program around less costly incentives (like honour, status and bragging rights) and allow individual teams to promise incentives for specific achievements (e.g. “If our council gets to number one this year, everyone gets £5 off their council tax bill”).
Coronavirus prevention is mandatory for all citizens.
Scoring programs can be designed on an opt-in basis if we want to achieve higher buy-in to the program. An example program which is currently run on opt-in is the test and trace apps that can be downloaded to smartphones but are not yet compulsory. There is the issue that the individuals whose social distancing behaviour is most likely to slip are also likely to be those that do not volunteer to install the smartphone apps. Of course the reverse may be true too (for example if having the app installed lulls users into a false sense of security) but it is probably a fair assumption.
Our program is going to be targeted at a group level (local authority area) so will be mandatory although since the data is aggregated it will be effectively anonymised, ensuring no data privacy issues.
Ok it’s now time to move to the Design phase.
First of all we’ll survey some of the options when thinking about what metrics we could use in our program.
We could calculate the percentage of shoppers in a specific town ( as in this article about Maidstone) complying with mask wearing guidelines. Apart from the fact that this is difficult for those who cannot wear masks for health reasons, it does provide a useful vanity metric. It allows us to compare shopping centres. If we could get the data for all shopping centres we could compare shopping centres on % of mask wearers. This would in turn encourage shopping centres to get very strict on mask wearing.
Any untimely death is one too many. In the same way any case of Covid-19 is one case too many. A design that focuses on these lag metrics (as typified by the current way public health officials share data) isn’t much use for a citizen trying to optimise their behaviour. It provides no signposts as to what they should or shouldn’t do, or should do better today. A lag metric only gives the results after the fact when it is too late to do anything about it. Instead we need metrics that give feedback on the effectiveness of social distancing and rewards players for maintaining these procedures. Lead metrics might be around mask wearing compliance, number of groups sticking to the rule of six, number of citizens self reporting they have self isolated and so on. These would provide valuable lead metrics that are something we can do something about.
One of the issues with Activity metrics (what I’ve done) is that they are easy to “game” - we just do more of something to “win”. While these do provide a useful signal in any gamification program, it is the reciprocity metrics, what others do as a result, that may be more useful. In Covid-19 we might imagine a system akin to Black Mirror’s famous social stars episode (Nosedive) where you are able to thank people who have passed you in a socially distanced manner. “Someone just passed you, how was their social distancing behaviour, good, acceptable, bad?” This may sound unworkable (and of course creepy) but you have to remember if the alternative is a full lock down, most people will be willing to adopt new behaviours if they give them freedom. However given the government's record with app design I wouldn’t pin my hopes on any metrics from an app this sophisticated anytime soon.
Mortality and case numbers, and indeed the R number, all fail on this design criteria. They are negative metrics yet we see them displayed as positive numbers. The problem is we are wired to try to increase the metrics we are given, no matter that they are mortality statistics. Here instead we should take a lead from the construction industry which has moved from “number of accidents per year” to “number of accident free days”. This is switching to a positive metric, one we can all get behind and increase. So for Covid-19 we could perhaps have a metric of “number of days since a new case” with the implication being that a region with the largest number of days since a new case in their community will be social distancing the best.
Clearly here we must focus on the metrics we need to get right first, which is probably why we hear a lot about the R number. It is the most critical management metric for handling an epidemic. However since R is a lag metric it’s difficult for us to optimise our behaviour directly for this. It’s not a very good gamified metric. We must find instead the metrics that are currently relevant to our community. And this is where a one size fits all program is unlikely to work - not all areas are experiencing the disease at the same rate, at the same time. Somewhere with high community transmission occurring will need to focus on a different metric from somewhere the virus is relatively dormant. The metric we focus on should be relevant to the timing and context.
There are many different behaviours that have been introduced as a result of this pandemic. While some we have got quite adept at quite quickly (replacing physical contact with virtual zoom meetings for example), some we are still at the stage of learning and adopting - mask wearing for many is still a matter of remembering to walk out the door in the morning, on a busy school day, with at least one mask in your back pocket.
Covid-19 has been spoken about with a daily rhythm - the government’s daily briefings, the daily release of new statistics. I think any gamification program would need to mirror this with a daily update of scores. In terms of score period this will depend on whether we are running a league - which might be all time or monthly. It will depend very much on the gamification design for the individual public health team and their local objectives.
While comparison may be the thief of joy it is often the catalyst that’s needed to attract attention - nobody likes being bottom of any table, especially if it is to do with Covid-19. Without a comparison of some kind your program may well provide useful feedback but will need a major communications push to get people to sit up and take notice.
This means that some sort of league table I think is inevitable. In this approach where we are trying to attract the attention of a large number of citizens, leaderboards are the lingua franca of journalists and football supporters the world over. Social media is awash with top 10s, top 100s and so on.
The key to good league table design is to ensure that no area is unduly disadvantaged. Built in inequity is a recipe for disengagement as players and teams feel they “cannot win”. For example, if a country leaderboard was ranked on number of deaths (as we see on the Johns Hopkins coronavirus dashboard) this is biased against countries with a larger population - a more equitable ranking might show deaths per 100,000 people for example.
Comparison also doesn’t mean necessarily sharing the whole comparison with everybody. For example a local council might generate a full leaderboard for all councils in the country on the metric that matters to them yet only publish the difference of their council against the average. Or, it could publish just how the council is doing against its neighbouring councils. In this way, it can focus citizen attention on driving good social distancing behaviour without getting bogged down in a statistics argument with councils far away that are not doing well on their leaderboard.
The best distribution channel for a public health official is probably to publish the gamified statistics on the coronavirus statistics web page in a straightforward way, and to then encourage journalists to add their own commentary. This is the approach of many sports bodies - the ATP Tennis rankings for example are displayed simply as statistics, yet these rankings are then republished and the changes discussed by many sports news outlets, each with its own audience - ESPN for example.
This is critical and really a vital role for politicians - those in charge of public health policy. Our program should be framed as a way to see how well we are doing on the metrics that matter because we want to be the best at social distancing. Getting the narrative right around what the program is for is really critical to its successful adoption by citizens. Labels are really important here - talking about social distancing, social distancers, covid secure and so on - are far more helpful than deaths and cases. No-one wants to be on a leaderboard of deaths even if you’re at the bottom of it.
It is not the job of this article to come up with a finished gamification design, I only hope that I have shown the ground that any public health design team would need to cover in their thinking.
Based on my own very brief analysis though, I think I would start with a leaderboard of councils updated daily with number of days since last community case transmission. I think that would help show the councils that are doing well at keeping covid out of their community and avoiding transmission. Those councils would be able to brag about something positive “we’ve managed 20 days without a covd case transmission” and crucially encourage their citizens to keep up the effort. It’s a little like the mechanic Snapchat uses for its Streaks feature: Teens explain the world of Snapchat's addictive streaks, where friendships live or die - if it took off, the gamification program would encourage communities not to let their guard down and maintain social distancing.
Here’s a mockup of how it might be displayed:
Now this is just an example final product, there are no shortcuts - you have to run the process with your own team that understands the data, your context and your current public health objectives.
After launching the program we can then expect it to evolve and iterate over time - the coronavirus environment is changing all the time, old rules expire, new rules come in. A good infinite gamification program though can evolve with the changing in rules and key metrics. For example if cases were coming thick and fast then the program design above might need to change to number of days without a Covid fatality.
I hope this article has shown you that:
providing statistics is often gamification even if you don’t realise it
Our public health officials can work harder at making Covid-19 and social distancing statistics more attention grabbing and actionable
Infinite gamification provides a design model for thinking through how to provide good feedback that people take note of and drives positive outcomes.
If you found this article useful then please:
- Subscribe to the Infinite Gamification Noise blog
- Buy the book
- Forward this article to your local public health officials or elected representatives