Weather forecasts and predictions for changes to the global climate have greatly improved over the last 20 years. Scientists have also improved at communicating these predictions, including any uncertainty. For example, a rain forecast may be assigned a 30 per cent chance to reflect the possibility that changing conditions could completely change a prognosis.
Forecasting is also critical to setting realistic development targets. The Millennium Development Goals (MDGs) aimed to, among other objectives, reduce child mortality by two-thirds, provide universal primary education and halve extreme poverty by 2015. These goals were meant to be ambitious, but realistic.
As we enter the final year for the MDGs, the latest UN reports suggest that only in some countries did reality live up to the goals.  Success stories, such as Brazil cutting child mortality by three-quarters, justified the ambition. But other countries, especially in Sub-Saharan Africa, will miss the goals despite making progress.
The question is, are the countries or the development goals to blame for the failures to deliver? We strongly believe it was the UN forecasts of what was possible, and hence the development goals, that were off target.
We have built a mathematical model based on individual countries’ development trajectories up until the time the MDGs were set in 2000, to forecast changes after that point.  It is possible to make these forecasts because very different countries have followed similar transitions over the last 50 years — from low to high levels of healthcare, for example.
Using data on progress up until 2000, we can say what would have been reasonable (at that point in time) to expect for 2015. The results show that many South American countries were likely to successfully reduce child mortality because they had a lower level of mortality to begin with, as well as stronger economies, compared with most African countries.
On the other hand, if Sub-Saharan African countries were to reach the MDG targets, they would have needed both unusually rapid economic growth and health improvements. That would have been unprecedented progress — much more rapid than that seen in Europe, Asia and South America in earlier decades.
“Numbers should not be the goals in themselves, but the tool by which we reach the ideals of the SDGs.”
The challenge now is to take what we have learnt and move towards realistic Sustainable Development Goals (SDGs). The UN’s latest proposal includes 17 goals that focus on poverty, the environment, social inclusion and economic opportunity — and the question for many is how to turn these ideals into numerical or percentage targets. 
Our answer is that we can’t.
Numbers quickly become irrelevant
To begin with, our model shows that goals have to account for differences between countries and continents — and the current SDG process makes no provisions for such differentiation.
The problems go deeper than this. Some goals are correlated with each other. For instance, lifting people out of extreme poverty provides immediate health benefits. Other goals can be contradictory. For example, investing in fossil fuel power plants can reduce poverty at low immediate cost, but pollutes the environment in the long term.
Setting arbitrary numbers by which to measure SDGs cannot possibly account for such real world complexities. In 1980, management consultancy McKinsey & Company forecast that the US would have 900,000 mobile phones by 2000. In reality, this benchmark was exceeded in 1987 and the actual number in 2000 was over 100 million. In our rapidly evolving world, numbers quickly become irrelevant.
Engaging the public
The pillars of development set by the SDGs — economic, environmental and social — are stable ideals upon which to build the future. And alongside that, there is a need to continuously engage the public on progress and remaining challenges in fulfilling these core ideals.
How can we do this? Firstly, as emphasised by the UN’s data revolution group, reliable data should be collected on topics such as agriculture, disease, technology and pollution. By making these publicly and easily available, the UN can ensure transparent evaluation of development.
Secondly, scientists will need to convert the large data sets into computer simulations of future development. Instead of a single target, such predictions should take the form of a range of desirable outcomes that include trade-offs. Using this approach, for example, reducing poverty by 40 per cent with a ten per cent increase in carbon emissions may be viewed as equivalent to a 20 per cent poverty cut with no emissions rise.
These trade-offs should be specific to countries, even regions within a country, and include multiple development indicators.
Predictions of development progress should be dynamic and interactive. It is often when weather forecasts are wrong that they generate the most interest, and the same should be true of development forecasts. For example, while Liberia, Senegal, Tanzania and many Sub-Saharan countries missed the MDG targets, they managed to reduce child mortality to levels below those our model predicted. Others, such as Sweden and the United States, underperformed relative to model predictions.
By accumulating data and comparing forecasts with outcomes, scientists can engage stakeholders in an active debate about what can be expected in the future. Setting targets for and pushing assessments into the distant future misses the point of understanding and communicating progress in development, which is to continually debate about how goals can be attained.
Numbers should not be the goals in themselves, but the tool by which we reach the ideals of the SDGs. Just as weather websites enable users to see weather patterns from different perspectives, modelling and visualisation should become a central part of communicating the core ideals of the SDGs and providing development forecasts that are continually updated.
We shouldn’t make the same mistake twice: unrealistic, static goals won’t work. We need to make a genuine attempt to communicate how predictions are made and progress is monitored. Anything less involves lying to ourselves.
David Sumpter is professor of applied mathematics at Uppsala University in Sweden. Shyam Ranganathan is a PhD student in applied mathematics and Ranjula Bali Swain is an associate professor in the department of economics, at the same university. Sumpter can be contacted at firstname.lastname@example.org and on Twitter @djtsumpter. For more information on his research, visit www.collective-behavior.com
References Levels and trends in child mortality — Report 2014 (UN Children’s Fund, 2014)
 Shyam Ranganathan Identifying the real success stories in development (Collective Behavior, 22 December 2014)
 Open Working Group proposal for Sustainable Development Goals (UN Department of Economic and Social Affairs, accessed 9 January 2015)