First, a pipe dream. 

 Episode 3: Through Muddy Waters.  Your fireteam is forced to move through the city center in order to hook up with Alpha squad.  Your Humvee is riddled and smoking, but it will hold -- it has to.  Expect heavy resistance. 

Co-op.  One player assumes control of the driver, maneuvering through the dusty city streets.  A friend is manning the .50 cal.  Two more friends are nervously panning their rifles through the Humvee windows.  Radio chatter indicates you must head North -- your driver swings around, looking at the compass on his HUD, and zooms North.  Mortar shells begin to THOOM THOOM too close to the Humvee, spraying dirt into the driver's face.  He fishtails but makes it to a narrow street and the shelling ceases.

Your two friends in the back decide to get out and walk behind the Humvee, while you drive slowly.  There could be landmines.  Or an ambush.  Your gunner swivels back and forth slowly, keeping his eye on windows.  Suddenly a pale blue pickup truck takes a corner nearby and picks up speed.  There is a heavy machinegun mounted on the bed.  Your buddies scramble for the doors and get in just as you hit the gas.  There was indeed an ambush up ahead, and bullets start to ping off the Humvee siding.  Everyone takes some hits, but your gunner is going crazy up top.  He rakes his .50 cal fire across a row of windows as you speed down the street.  The guys in the back are laughing out loud over voice chat.  Forty five minutes later they make it to Alpha squad; the screen fades to black and promotes next month's episode.

Television episodes are not dissimilar to game missions.  They operate in serial format; each mission might have the same cast of characters, but there are new events.  So what's the problem?

Current developers can't just switch over to an episodic production model; like Valve perhaps they can change the way in which they produce their normal games, but Valve cannot turn into Telltale and remain Valve.

So, episodic games require an episodic-centric studio -- new studios with the intent to produce episodically.

They also need a platform -- I think it's safe to say that Steam is that platform. 

Once my cousin's husband asked me if this was possible: A Grand Theft Auto game with a fully-developed racing model and a more involved shooting system.  I told him that would be awesome, but it would take forever to produce -- long enough to be unrealistic.

Because games with four year development cycles can only afford (literally) to focus on one thing.  Special games like GTA can afford to focus somewhat on several.  It's unlikely that the dream game above could have a serious driving model developed for it, or that the gun combat would be impressive.  By adding more genres together you dilute them by denying each of them development time.

The only answer for that is advanced middleware support.  If a company like Turn 10 were able to wrap their racing model up in a friendly package, or pay another company to do so, and license it, then things would be easier.  It would still be rough going early on at an episodic development studio as everyone learned how to handle the integrated middleware, but it would make development times realistic.  It would also provide companies that finished the sort of system that such a studio might need some extra profit -- we've all seen the wonders licensing an entire graphics engine has done for Epic.

So I don't think episodic gaming as imagined above will ever be realized, not unless those stars align.  And I'm a pragmatist (call me a pessimist), I don't believe they ever will, or not in such a clean and benevolent fashion.

So maybe episodic gaming has to be different -- maybe the popularity of DLC is a sign.  But I'm going to continue to imagine that original content in episodic form could be possible, if improbable.