Notice my use of the word both. In this relationship-centric game, the A.I. has personality and sexual preferences. Ensuring their needs are met reinforces the ideas of consideration and concern for the other's well-being. From a gameplay standpoint, this introduces an element of uncertainty to each new encounter, since each partner will have his or her own unique preferences.
From an interface standpoint, movements should imply the real action as closely as possible. Variably sped rotation along with forward and backward movement can all be controlled using an analog stick. Individual positions could rely on unique combinations of speed, rhythm and accuracy. For obvious reasons, a vibrating controller is a must.
Of course, pacing the encounter is an immediate concern: You must indicate when both participants are satisfied. One option is to use visual and auditory "excitement levels," which both partners could display through a combination of animations and speech. Additionally, this would allow the A.I. to provide direction to the player. Switch to your partner's desired position without the need for the A.I. to request it, and you raise your partner's excitement level. This system allows for learned skill after repeated attempts; over time, vocal queues should not be necessary, and the player's knowledge of the other person's preferences would become his or her sole roadmap.
Since the goal is ensuring pleasure for both characters, the game would be as much about controlling your own excitement level as it is reaching the height of it. That could mean short pauses or changing speed once your partner has reached his or her optimal state. If either party reaches their end goal much sooner than their partner, it could become more difficult or impossible to continue, depending on certain - ahem - biological concerns.
In this game, "grinding" could take on a whole new meaning. Every sexual encounter could potentially open up new maneuvers and techniques based on the A.I.'s positional suggestions. Over time, the player's collective experience upgrades their options for future encounters.
That's not to say there aren't major problems with this approach. Firstly, there is no feasible way to create animations that convey all of the subtleties in people's facial expressions, body language and vocal reactions. An amorous picnic in a quiet park could become a harrowing excursion into the uncanny valley. But on a more basic level, gameplay takes imagination out of the equation, replacing it with button presses that detract from what the focus should be.
Maybe there's a compelling reason for static sex scenes after all. They bypass the uncanny valley by showing a few vague visuals, and they take away control so that the player can relax and enjoy the fruits of his labor. Digitizing sex while attempting to keep it realistic is a difficult and contradictory endeavor. But experimenting with the idea can lead to interesting A.I. and valuable attempts to convince the public that sex needn't be equated to pornography in games.
Michael Deneen is a lucky designer.