Back in the old days of D&D, that was how your character was created. 3D6 created ability scores, from a range of 3-18, with 10 being defined as “average.” In the most hardcore tradition, you rolled six sets of scores and assigned them to your character in the order in which they appeared, giving you a completely random character. Then you’d pick a class your character might qualify for (with a ‘fighter’ — or ‘fighting-man’ — really having no qualifications necessary) and off you go. If your character was truly pathetic, you might hope to have ‘em die quickly so you could create another character. Creating a new character in old-school D&D took only a few minutes, so it wasn’t a big deal.
At some point, people decided that those who took the adventuring lifestyle ought to be at least somewhat better than your average pig-keeper (never mind that heroic fantasy does have a place for even assistant pig-keepers), and opted for more generous probabilities, and the ability for a player to choose which scores went to which abilities. In the AD&D days, the preferred method was to roll four six-sided dice for each score, but to ignore the value of the lowest die. This still yielded scores in the 3-18 range, but with a higher average, and it was still possible to get a really weak score in one or two abilities.
D&D – and most other game systems – eventually moved away from randomized stats in favor of “point buy” systems. Players no longer needed to fear a bad set of dice rolls! While it sounds great on the surface, the problem is that player characters all end up with very similar sets of stats, min-maxed for their chosen specialty or class (and if it’s a classless system, it’s even worse).