Stable and unequivocal references to datasets, samples, …
In most cases, the number of items we deal with in our research – be it samples, calculations, measurements and alike – exceed by far what we can handle without any kind of system. Honestly, even more than ten different items usually are too many for humans to handle properly without a clear “naming scheme”.
Furthermore, analysis of data usually nearly always requires comparing multiple datasets. Hence, we need to be able to provide (unique) references to those items, be it datasets, samples, or else.
For references to the literature, such a system has been established centuries ago and has helped tremendously to cope with the ever increasing body of literature. Furthermore, this referencing system has been updated for the digital age, using “digital object identifiers” (DOIs) that are nowadays not only used for research articles (and sometimes book chapters), but increasingly as well for other digital items.