# Accuracy is a Myth By:: [[Brian Heath]] 2022-11-18 Accuracy is a delusional narrative rooted in existential [[anxiety]]. It has nothing to do with reality and the nature of the universe. Let's say an analyst presents a statistical finding and states they are 95% confident that the value will be between 15 and 20. How would you characterize the accuracy of this finding? First, it's impossible to say as you do not have any context about the system. A value of 15 may represent survival, and 20 represents death. However, the most important part of the system context is how you feel about it. If you don't care much about the number, you might say this is a very accurate prediction. If you care whether the number is 17.2 or 17.5, you will likely find this prediction entirely inaccurate. These statistical confidence intervals that proclaim a level of accuracy provide little comfort in the face of your contextual reality. That is to say, how you view the idea of accuracy depends on your state of mind. There is no reality in accuracy - it's just a thought that lives in your head. To further illustrate the point, let's unpack what is meant by the statement that the analyst is 95% confident in a range of numbers. What does 95% confidence really mean? From a statistical perspective, this statement means that 95 out of 100 experiments result in an outcome within the range provided by the analyst. Maybe that gives you some comfort about the accuracy until you realize that your problem isn't a lab experiment. You only get one shot at most things in life. So, are you willing to gamble it all that this one time will be part of the 95 or 5? However, this isn't the most critical question of the moment. If a series of events can only ever happen once, where does the analyst get off claiming 95% confidence? They can never be proven right or wrong. This confidence range exists only in the world of thoughts and simulation. Only once the thing happens can any assessment be made about the confidence statement. However, the analyst has an out either way. If the outcome falls outside the range, you are unlucky. If it falls within the range, the model is indeed accurate. Thus, you see that accuracy is not a real thing, and only a narrative to express our level of comfort with the uncertainty ahead of us. Reality will always be accurate to reality. Everything else is a delusion and distortion. It just so happens that some [[Delusions]] and distortions are more useful than others (e.g. all models are wrong, but some are useful). From this perspective, the most accurate statement an analyst could give about the future is it exists until it doesn't. #### Related Items [[Analytics]] [[Predictive Analytics]] [[Forecasting]] [[Accuracy]] [[Confidence]] [[Statistics]]