While this phrase is intended to apply to all fields of science, I believe it is particularly pertinent to where we are now in the application of Systems methods to immunology. It’s also a good reminder that all science starts with observation. This goes against common wisdom, which holds that one should start with a hypothesis and work their way to’mechanistic’ facts by the conclusion of the paper. We certainly want to focus on hypotheses and mechanisms, but thorough observation and analysis must come first. Hypotheses are only worth having as additional data accumulates. I’m biassed, but I believe the modern form of Systems Immunology began in 2008–2009, with a relatively concurrent publication by me titled “A Prescription for Human Immunology” and the first data papers by Sekaly and Pulendran, where both groups used gene array data and other data to analyse Yellow Fever Vaccine responses. The common thread running through them all was that we required new methods to human immunology, because so much of what we do in mice can’t be replicated in humans. We also wanted data that wasn’t reliant on what we know about mouse immunology, because numerous failed translational efforts have shown that the mouse isn’t a good predictor of human outcomes