Though I understand that espionage can be more of an annoyance rather than an interesting mechanic for most singleplayer players, who just like building up their empire - as someone who plays mostly multiplayer, I hope it will not be downgraded to a low impact side mechanic (send and forget, as some describe it).
For multiplayer, espionage, if implemented well, is both a great catch-up mechanic and a fun way of harrassing your opponents. In fact, some multiplayer games suffer from a lack of interaction other than the occasional war or World Congress vote, and espionage could be a great system to help mitigate that.
I accept that the majority here plays mostly singleplayer and therefore holds the opposite view. That said, I would like to advocate for a middle road for the above explained reasons.
I mostly play SP lately, but I agree it should impactful -- every MP game I've played has nonetheless had more AI than human so it must also suit the AI, but the more that can be accomplished with spies, in a way that makes sense to civ/VP abstractions, the better
In seriousness, while we have all done our best on the various drafts, they are still drafts. Espionage is a complex topic, and as this discussion shows, we have a lot of diverse opinions. More than anything we need feedback to try and do some polish before we release any of these ideas into the wild.
Two of the draft proposals seek to connect spy travel to the gameworld, albeit in simplistic ways. I can't remember which, but one proposes on the basis of technologies, the other one city infrastructure/connections. These are steps in the right direction, but too little.
To remove the spreadsheet feel of civ 5's espionage, the spies need to actually move in some fashion that reflects the map. We cannot currently support them being units civ 2 style, and this will realistically never be possible in civ 5 short of retrofitting some kind of ChatGPT machine learning AI into the engine -- this is not what I'm thinking, but the distance from home capital, or their present location if deployed, should matter to their movement, especially early on when spies are presumably travelling on foot/by sailboat. Just spitballing here, and this would require some in-depth coding to develop the backend logic (though not unattainable, i'd argue), but consider that at present due to city working radius etc., its highly unlikely that cities will be more than 6 plots apart -- so give spies a 1-turn move radius of 6 (or 7 as failsafe), and when player sends spy to a far-off city, it actually must transit through the other cities to get there.
Eg. London is 30 tiles from Paris, and the land-grab phase is over, (nearly) all possible cities have been founded: english spy has to get to paris from london, finding intermediary cities no more than 7 plots away on each turn => travel time will be 5-6 turns, roughly in realm of status quo, and we have no AI-breaking spy unit, just the current system more or less, but with spy actually changing location in a way that fits the in-game world. After Liz gives orders to go to Paris, english spy ends turn 1 in Leeds, Turn 2 in Birmingham, turn 3 in Avignon, turn 4 in Marseilles, turn 5 in Paris, etc.
A big part of the counter-spy game could then become trying to catch spies in transit, secondary cities become relevant, events in game-world could affect the city-leapfrog "route" they must take, etc. As time goes on, this movement restriction can be relaxed, the city-leapfrog range increased etc. Much of this could still be automated, fire-and-forget type spy instruction
unless some intervening event takes place
Anyway, first proposal that attempts to give spy some kind of map-connected movement will likely get my vote.