Today we recognize the central role that intrinsically disordered proteins (IDPs) play in cellular processes. This was not always the case. To understand why the IDP concept had not surfaced previously, we need to take ourselves back to the scientific mindset before 1990. The human genome project was not on anybody’s radar – very few if any genes had been sequenced. Cloning and expression of labeled proteins was in its infancy. Computing power was pitiful compared to today – the capacity of our institute-wide mainframe was less than that of a smartphone today. For protein NMR, the dominant technique was all-proton, and 2D was only introduced in 1977. Heteronuclear spectra of uniformly-labeled proteins were introduced in 1989. Our lab was focused on the study of peptides and small proteins, which seemed somehow qualitatively different – peptides were invariably disordered, sometimes with a propensity detectable by NMR for conformational preferences for local structure, whereas proteins were ordered in three dimensions, with a single overall structure that could be reflected in an X-ray crystal structure or, excitingly, by NMR. During the early 1990s we became increasingly puzzled by a series of proteins whose sequences were derived from the earliest sequenced genes. The genetic information was unequivocal on the location of the functional regions of these proteins, but when we looked at them in the NMR, they behaved like peptides. These proteins were disordered, yet still fully functional. Many of them fold into ordered structures upon binding to a partner; many remain fully or partly disordered even in partner complexes. In asking ourselves why disorder might be preferred over order in some types of proteins, we realized that disorder confers important advantages to the efficient operation of cells. Recognition of IDPs would not have been possible without the unique insights provided by NMR.