Contemporary wireless traffic is increasingly heterogeneous, with growth coming primarily from low-rate unattended devices. Specifically, whereas human users tended to establish sustained connections, machines transmit succinct packets sporadically. This represents a formidable challenge for (a) the initial access in the licensed-spectrum networks (LTE 5G etc), and (b) the dedicated IoT long-range unlicensed networks (LoRa). This demand for a redesign of the medium access layer, prompted an exciting set of new information-theoretic and coding-theoretic problems, named unsourced/uncoordinated multiple access (UMAC). Correspondingly, in this tutorial, we discuss both the fundamental characterizations and pragmatic algorithms. Excitingly, UMAC can be interpreted as a very high-dimensional (2^100 and more) compressed-sensing (CS) problem. Thus, UMAC theory and algorithms are equally applicable to domains where such CS problems arise, for example in estimating heavy-hitters in data streams. For the latter problem we will show how communication-inspired solutions compare to computer-science algorithms of count-sketch family.