I have dabbled in freeform. It works. This begs the question; why use rules at all?
There have, naturally, been several good answers to the question. One famous comes from the Forge and is called Lumpley principle: “System (including but not limited to ‘the rules‘) is defined as the means by which the group agrees to imagined events during play.” In essence: System, including the rules, tells who can say what and when (“what” regarding the fiction). So, what are the implications? First, all games with one GM and number of players who only decide things about the behaviour of their characters are fundamentally the same. Second, there are alternatives to that model. Third, there is always a system; rules just make it explicit.
Another, more recently surfaced, theory says that rules share spotlight (originally from Ben Robbins, further elaborated by Fang Langford here, here and here). The implications include that rules which create separate subgames (hacking in Cyberpunk) not accessible to all players are bad. In turn-based environment, such as many combat systems, rules that give extra actions should be considered carefully. Critical hits and fumbles are actually meaningful rules; they give random bursts of spotlight to players. The purpose of rules is to regulate spotlight away from those who would have it in freeform gaming.
Rules also create shared expectations with regards to the fiction and player behaviour generally, which is probably more important. The Mountain Witch (Vuoren velho in Finnish) has rules for betrayal and every player knows that every other player has a dark secret. This creates the atmosphere where betrayal is assumed to happen, at some point, and hence reduces the potential for damage that may happen with unexpected interparty conflict (see also: this RPGsite thread). Game art and such also contribute in similar manner. Under this point of view rules make the gameplay smoother; less need for negotiation, fewer mistakes that need fixing.
“The rules are the physics of the game world.” is often heard. I am going to extend it a bit: Rules define setting. The effect is not necessarily straightforward, but it is there. D&D dwarves are harder to hit by giants. There might be several reasons for this, or none at all, but it will be reflected in the setting somehow. What does this imply? Well, first, the more rules, the more defined the setting will be. This may be a perk or a flaw. Second; setting andrules interact. Choosing one will or should affect the other. Third, if you want a setting that does make sense internally, take a good look at the rules and how to translate them into in-game (diegetic) information.
Rules create tactical and strategic sub-games. Some are very explicit about this, some far less so, but at least the combat or generic conflict resolution systems tend to be minigames. Implications: Make them meaningful and fun, given that they are there.
In summary, there are several ways of looking at the purpose of rules. When selecting a ruleset, or designing one, it is useful to look at it from several perspectives.