Projects of Self & Risk Forecasting

I work for (in my opinion) a fascinating company that uses machine learning to forecast risk and outcomes on major construction projects. It’s essentially predicting the future and I think that’s pretty cool. 

Granted, a lot of what I do involves learning technical information so niche that I cannot share it or discuss it in any other area of my existence. I have, for instance, been at parties - three pints down - trying to explain critical path analysis, PERT distribution models and graph neural networks to strangers and friends. Often, and understandably, this is met with blankness, disinterest or pity. This is because it’s hardly relatable content until I flip the narrative a bit and start to discuss risk more generally: how do we go about forecasting risk in our everyday lives? What data do we use from our experiences to assess that risk? Do some people’s datapoints leave them more vulnerable to not recognising risk and how do we change those datapoints if they are core experiences or core memories? Are some people just naturally inclined towards risk or is it a product of nurture? How do we process data to make those decisions when that data is so complex and how do we present that information to other people? etc etc

My answer to the above involves drawing a comparison between construction projects, our lives and storytelling/writing. The definition of a construction project is 'the organised process of constructing, renovating, retrofitting or adapting that results in the creation of a tangible product or outcome.' A story or a piece of writing isn't so dissimilar - it involves a process of organising certain ideas and granular elements into an overarching permanent structure that has a desired impact/effect. Then there's our lives - well those are always in a state of permanent construction (and deconstruction) as we constantly adapt, evolve and recreate our identities or narratives of 'self' against the backdrop of time. The analogy of a life as an unfolding story is age old but what about the analogy of a life as a construction project? Both have a start date and a finish date with a shit tonne of activity in between that aggregates to form the overarching narrative of how that project turned out. Without labouring the point, I think there's some pretty strong interrelation between the three.

But now I want to talk about where risk and uncertainty fit into this equation.

When I think about the transference of risk within construction to risk within my life, it makes me think a lot about my relationships and friendships. How many times have I risked being vulnerable to people who I shouldn’t have been vulnerable with? How much risk does trust carry? When is risk worth it, when is it detrimental and how thin is the boundary between the two? How many times have I said "I'm absolutely sure this will happen" only for the opposite to happen. How many times have the safest most secure things become unsafe, unsure, uncertain? etc etc

It’s funny that I’ve never really seen friendship as a risk, even though the acts of trusting and loving someone are inherently risky. Only a few times in my life, due to some pretty shoddy behaviour, have I felt a friendship to be a genuine risk to my own happiness. Most the time, the love feels pretty simple and therefore involving very little risk or threat. 

On the other hand, I think romantic love is always a significant risk and a massive leap of faith. This is because romantic relationships are similar to nuclear energy projects - bloody complex, long-term investments, often large-scale and could blow up at any point..(technically not quite accurate from a project perspective but roll with it) It is worth noting that nuclear energy projects such as Sizewell-C and Hinkley Point-C are also what will fuel our future and sustain us. No pun intended but they are literally energising and seriously bloody important. Romantic relationships can also be incredibly energising and rewarding but that is not to dismiss the serious amount of risk involved along the way. As with anything risky, romantic relationships should be handled carefully and with consideration. I think too often we jump in without thinking about the implications or inherent risks the relationship might pose to our own happiness and the other person’s happiness. Learning to handle ourselves and other people with care in those situations is therefore the ultimate key to de-risking such relationships, and ensuring both parties remain a source of opportunity/reward to one another.

In the construction industry, traditional risk analysis methods rely on the limited and finite experiences of a very small number of people. They will say “based on my experience of the 5 mega-projects I’ve done in the past, I think this project will take X number of years and A,B,C,D are likely to be the biggest risks”. It’s very subjective, and clearly pretty ineffective considering that 6/7 construction projects are delayed by at least a year. Essentially, poor data means poor decisions are made and risks go unforeseen only to materialise later as seriously costly issues. 

So I must ask, how many times in your life have you been in a situation where you’ve been blindsided by how monumentally fucked you are once its already too late and you’re in too deep? What if you’d just had some foresight? What if you’d just sat down and thought more carefully about the risk involved in the decision you made or the path that you chose? What if you’d just used the data you had in front of you to do something different with your future? What if you’d had access to more experiences or more data to change your decision-making process? 

This is why my company exists - essentially, we use machine learning algorithms trained on a dataset of 760,000 past project schedules to accurately forecast risk on complex mega-projects. So the hypothesis is - the more data you have, the more you can learn from and the better your prediction of the future will be. Essentially, it’s providing a data-driven approach to risk management that means more informed, strategic and proactive decisions can be made.  

So if, by analogy’s sake, I am one massive construction project then the nPlan to my life would involve accessing the lives and psychological trajectories of 760,000 other people to flag up to me when my heart would likely get broken or which job path would carry the most risk or if I said XYZ to someone this would likely be their reaction. God, life would be so simple. But unfortunately I don’t have that and we never can, because we can’t access people’s lives in the same way that we can access project schedules. All I have is my own inner life, my own thoughts, my own experiences and my observations of the lives of others. However, that in itself forms a pretty large dataset. 

Now, I’d love to say that knowledge is the data-point that informs most of my decisions but very often I am driven by my emotions. I will “know” that something isn’t good for me or that it carries risk but do it anyway. This is because my emotional/psychological make-up, like every other person, has been influenced by what I have experienced, both in my childhood and adulthood. The result is a clusterfuck of unprocessed, unstructured and sometimes conflicting data.  

Therapy has been a brilliant way of processing and structuring that data so that I can make better quality decisions that carry less risk or detriment to my life. For example, I have traditionally really struggled with walking away from things I know to be bad or hurtful for me. I have also really struggled to distinguish at what point a risk is good or bad and this is because I find it difficult to differentiate between that which is genuinely exciting and that which is anxiety-invoking.  I now understand why I made those decisions and can make sense of the data that drove them ie X data from my childhood makes me inclined to people-please, but Y data from my adulthood has taught me that making decisions based off X data gets me hurt because X data was collected from an already hurtful, harmful situation/dynamic. Using Y data, I can therefore better understand X data, and also know that Y data should be used to flag up risk in future situations. Essentially, X data helps me to recognise similarly toxic patterns in relationships whilst Y data lets me know to walk away. This has now enabled me to more easily spot risk in relationships, avoid those risks and understand that those risks aren't/weren’t always my fault. I think this has been a real game-changer for me and has removed a lot of risk in my life that I otherwise would have accommodated. 

What’s also interesting about the construction industry is that a vast number of rail projects will go wrong and continue to wrong. You’d think after 10 delayed rail projects, someone would say “wait a second, something we’re doing obviously isn’t working”. You’d think after 100 delayed rail projects, someone would say “what the actual fuck is going on, what did we get wrong and how can we avoid doing that again”. Essentially, the same mistakes get repeated continuously and whilst there’s a number of reasons for why this happens, it is often because risk analysis is influenced by optimism bias, stakeholder bias or recency bias. So instead of using raw data to predict what will most likely happen (ie observing what was planned vs what actually happened to guide how that might play out in the future), they let external factors influence their decision making. This leads to skewed results and repeated patterns. 

The same thing happens in our lives. I can’t tell you how many times I’ve said “well I’m never doing that again” only to do it again. But it’s WHY we keep doing things again that matters, and so often this is because of psychological biases. 

How many times have you let other people influence your decisions? (Oh they’re probably right this time, oh but they’re my best friend, oh but they’re more senior than me so I'll let them dictate this decision even though I know it to be wrong - that’s stakeholder bias) How many times have recent situations influenced how you’ve interacted with people in current situations (that’s recency bias - ie my recent ex did XYZ, therefore I am going to approach you as if you are also going to do XYZ even though there’s no evidence that you will repeat this pattern of behaviour) and how many times have you said “no this is different and I’m sure this time it’ll be better” (that’s optimism bias - they are showing all the same red flags as the last person and still you convince yourself this will end better - honey, it won’t)

I know that my ability to forecast how ‘risky’ a person is to my life is fundamentally skewed by optimism bias. I tend to see the best in people without actually acknowledging the facticity of their actions and how their past/current behaviours might allow me to predict how realistically ‘safe’ a person is to have in my life. If I relied more on data, I’d have a much more objective view of a person’s risk profile. So now before jumping to “god they’re the best! I trust them so much! I love them so much” I’ll ask myself a series of questions:

How is this person showing up for me? How do they communicate? How do I feel when I’m with them/not with them? How do they react when their behaviour is challenged? How do they treat waitressing staff? How do they treat my friends? How do they handle conflict? Do they ask me questions? Do they take an active interest in my life? How do they show they care about me? How have they treated past friends? Do they have a good group of people around them, if not then why not? How do they talk about other people? What is it that makes them so “fantastic”?

What this does is create a mesh of datapoints that allows me to take a more objective view of a person that sits outside of how I might feel about them.

So I've started to analyse my dataset better but with what tool? My company has a machine learning algorithm to process that data. I believe that I have writing, as it is only through writing that I can start to organise that data into something genuinely insightful. And so we're back to the tripartite link between lives, construction projects and literature. It is only through writing that I have been able to understand the huge amount of data that sits in my head. It is writing that has given my unstructured, sporadic life a real sense of direction and narrative when it needed it. In the same way that my company provides clients with a narrative that helps them know where opportunities and risks lie in their construction projects, writing has provided me with a narrative that has helped shape my thinking and the way in which I make decisions.