A lot of good ideas already, I'll just recap what I personally like.

1) Diplomacy: I love being able to sell almost/trade almost any resources. Being able to become an arms dealer is tons of fun. The dialog should be unique and witty for each race.  They should be able to tell if a trade is good or bad, and compare that to how badly they need it, as well as remember if I try to consistently rip them off, or shower them with gifts.   First Contact with dialog and video in GalCivII was always a blast, it must make a return.   I should be able to keep a trade agreement at a certain level without ruining every single relationship (some races getting pissed I don't want to invest further is understandable).  The longer the alliance, the worse the breakup, but I should be able to slowly decouple from the alliance.  Espionage (stealing tech, damaging cities, protecting cities, setting up roits, overthrows, etc).  After enough options to really create some interesting galactic scenarios, there is one other very important aspect.

Each race must be unique with a unique personality, though this personality should either be able to be set to random, or to a specific race (Ie: Humans that behave just like Drengin, Drengins like Iridium). I believe GalCiv2 had this capability either through UI or modding. 

2) Turns:  It all depends, honestly I felt that GalCiv2 was setup perfectly as far as turns go. When I wanted a shorter game, I'd pick a smaller galaxy. Some of my favorite games though took weeks to complete in full. Sorry, guessing at a number of turns would be like guessing at a jar full of jelly beans, but at times 500+.

3) Good Strategy Games: All good games had: Lots of customization/randomization. Specifically when it came to deciding how another race/culture would act. Random and multiple sized maps.  The game threw curve balls from time to time, but not enough to be excessive. Learning to late that another species has been behind the recent hostility of another race is an interesting scenario to come across. Being able to sell arms to two races on game and insight a war between the two is always fun. However, I shouldn't be able to do that every single game when I see those two races.  I had to have the ability to play a random game that felt like it had it's own unique story taking place. That means the AI had to be intelligent enough to vary it's approach. not always do the same play, no matter how stupid or powerful that play maybe. 

There's so much potential with diplomacy, what was in GalCiv2 could only be the beginning. I like being able to persuade them to do things that are not in their best long-term interests, but are in mine. Persuade a civ to focus research and/or pursue (avoid) specific research trees. Open border treaties so I gain some influence and/or citizens for soldiering on planets I don't own. Maybe I can influence local social issues on a per planet basis (environmentalism to reduce manufacturing focus, help overthrow leadership so the planets or the whole civ changes their strategies, help quell revolts). And my persuasion ability should be affected by not only diplomacy but might (military, wealth, production, influence, approval, ideology, etc.) too And other relationships (friend of my friend).

I loved how in MoO you could frame others for espionage and get them mad at each other. (and it was infuriating, but in a good way, when it happened to you)

1) It might be best to discuss what features I hate most.  While I love the Total War games, I'm always annoyed at how gung-ho the AI is to declare war on me.  Their system is disappointing because it can't affect meaningful impact.  I like it when diplomacy opens meaningful interaction with the AI faction.  Civ V also kind of fails on this front; the AI is only out to win and won't support any proposition that directly furthers a player's bid for a Diplomatic Victory. 

    I rather liked the old Vassal system Civ IV had.  It offered a means for a civilization to capitulate without being destroyed, and even as a vassal it had the potential to regain sovereign status if it performed well enough (or your civilization performed poorly enough).  If I had to pinpoint a single most important feature for a diplomatic system, I think it would be a meaningful capitulation system, something that can circumvent a long and drawn-out process of conquest when there is clearly no actual contest left to the fight.  Extra points if the capitulation system adds new facets to the relationship that can be developed upon.

   Even with that, I think the most important part of a diplomacy system should be a sense of character.  I should know what the other civilizations think, they should have agendas, strategies, and plans that can be related to the player with sufficient investment.  I should be able to cultivate a relationship with the AI.  There should be ways for us to interact where we can both profit, and if an AI has grievences against me I should be able to relate (which is to say, it should be for understandable reasons that I may or may not actively be aware of... not something abstract like a value that deteriorates with proximity or some other artificial construct). 

2) I really couldn't tell you a definite turn count.  I prefer longer games, to be sure.  I think an average Civ V game has good pacing and a satisfying turn count around the present day, near-future point, but I'd be happy with a longer game if it could still be engaging.  A game could go on ten times as long as Civ V if it still felt dynamic, if it still felt like I had meaningful choices to make that would affect an outcome that is far from certain. 

    More importantly, though, I hate it when 4x games have a hard turn limit.  Shogun 2: Total War is a perfect example; you have about 35 years to handle the main campaign, and that looming deadline puts an unpleasant shadow over the rest of the game. 

3) This is a complicated question. 

    A big pitfall for 4x strategy games is game-play diversity.  All 4x games are composed of a handful of components; city/world building, exploration, combat, research, etc.  Quite a few games have had mostly solid aspects, but failed with one aspect, and that failure dragged down the entire game.  This happens a lot with elaborate combat aspects, but any major system that is significantly dysfunctional can derail a game.

    This is likely a personal thing, but most of my happiest memories of 4x games have to do with the early turns and expansion.  This is less true with Gal Civ II, but defining the core shape of my empire is usually my favorite part of a Civ game. 

1.  I want to trade maps.  I hate the warmonger penalties in Civ 5.  If I get attacked I shouldn't be penalized for winning the war. 

2. It's a long mop up phase that is tedious.   The end game is usually boring and long t2urn times for AI to process.  I like about 12 hours of real time.

3.  My favorite part of gal civ2 was the scramble for planets.  

God yes.   Being penalized for wiping someone out who attacked you is BS and I hate how long the penalty lasts.  I mean I can conquer someone in 0 AD and everyone still hates my guts in the year 2000 AD.

Making other civs temporarily leery of me is one thing; but forcing me down the path of being an international pariah for all time because I conquered one annoying civ really undermined the diplomacy in Civ 5.

SE5's diplomacy amused me.  But I found the SE* series so unchallenging that I always play Team Mode, all humans (i.e. me alone) vs. all AIs.  Then I set "many" AIs on "hard" difficulty, and hope for 17-19 of them.  (If I get "only" 12, I restart the game.)  They team up on contact, trade techs, all sabo me with intel from afar, and tag-team me with streams of their entire production.  And I still kill them faster than they can get to me, so I never see unified fleet furballs like in SE3.  I like to stare at the empire scores graphs, I can tell when each AI meets me because that's the downward bend in its graphs as they all plunge toward 0.

I only ever played Captain Kwok's Balance Mod, so I dunno if the diplomacy is his or original.  Basically, I ignore every AI's 1st request, promptly respond by declaring war, and ignore it thereafter.  Heck, Team Mode already means they're all allied with each other and at war with me.

I actually won one game of SE3, all the way through the mop-up phase, until I conquered the last AI's last planet.  The end-of-game victory splash was ... one tiny pop-up window, roughly half the size of the "Post Reply" button at the bottom of this editor, saying You won! with an OK button.  I've long suspected that SE3 was written in Visual Basic.

In SE5, I've never lasted longer than about turn 150, or earlier when I hit the midgame Stolen Colonizer Tri-plosion phase (where my boarding party frigates have stolen one of each colonizer type I don't already have, and suddenly I can colonize all of the equal number of ice and gas planets throughout my sprawling backfield -- this could be 100 extra planets). Micromanaging in SE5 means turn time increases roughly linearly with things to do, until every turn takes me 60-90 minutes (not kidding). Somewhat before then, I always realize that the game simply has become not fun, and doesn't captivate me.  Every planet feels the same, every tactical ship combat is the same, it's just more of the same.  (I loves tactical combat, but even I can't stomach that much of it.)  Even when you can research and build a class-100 ringworld (their planets are hexless lists, but otherwise similar to here: tiny 5, small 10, medium 15, large/homeworld 20, huge 25 -- so ringworlds are supposed to be a Big Deal), there just isn't enough variety in the ground facilities to build thereupon.

The SE* series caused me to think deeply about gaming UI, which I'm still mulling.  Roughly, my grand idea is that a UI (and not just for games) must adapt to scale up to the complexity of its environment.  As the scope of your thinking increases, the tools at your fingertips, and the level of detail (LOD) you see, also must increase.  Hence the vocabulary itself evolves.

• One violinist knows what it's like to drill for one note at a time, with staccato, vibrato, etc.
• At conductor-level, you think in sections, phrases, timbre, inflection ... your vocabulary isn't even single notes any more.
• At composer-level, you might think even higher, in genres, moods, emotions.
• At orchestra-level, you manage buildings, storage, calendars, repertoires, guest itineraries, talent searches, etc.
• At conservation-level, you procure materials for musical instruments, even to planting groves of pernambuco trees to ensure wood for bows for 50 years.

Complexity is hierarchical (or we seem to make it so, because we can't cope otherwise).  Management also becomes hierarchical: we chunk modules into self-contained units, maybe with one vice-president to speak upward on behalf of each (and flog downward to meet goals).  Zooming out to higher LODs, we sweep the same human gaze across a vista that must be capped at similar total complexity ... but with the grain size now magnified, so that every node / pixel / unit encapsulates a complex state of its own.  What do you look for, and how can you be sure you're not missing anything?

This seems to require automation, but the player needs more control than a canned set of ministers.  (Partially, I lean this way because I think I can script code better than any stock writer-of-ministers, and I'm willing to compete vs. other scripters).  Maybe the first thing we do is embed Common Lisp (because we know it ends up in there somewhere anyways   SE5, and every 4x I've seen (including, ahem, GCs 2 and 3), all suffer from the click-only UI: what you do on turn 1 by clicking one pixel at a time is the same thing you do in turn 200, just more of it.  There is never an upward-shift, or lifting-of-level, where you define your own theater / unit / department, tell it how it shall run itself and report to you, and then let it go and trust that it works the way you want it to.  GC2's ministers were useless to me because in 10 minutes, I determined that I shall play the game a certain way, and ... they didn't.  SE*'s ministers are similarly shallow-in-scope.  N.B. an AI is just a null player with all ministers turned on, which is proof that the ministers aren't enough.

• SE* all made rendezvous important (e.g. colonizers to colonizables), but had no UI support for it beyond the click-level order.
• SE* ground facilities had scope: mostly planet-only, but some were system-wide, some were empire-wide.  Some are pop-dependent (so you migrate them onto your high-pop planets); some are pop-independent (so you migrate them to the tiny/small).
• Hence, by midgame, every system (which may contain 2-15 planets) must optimize a set of "1 of each of these", plus each planet is a queue.  How do you micro-manage that?  You'd need something like a tree-shaped UI that inherently shows you "1-per-system" items separate from planet queues, and all queues at once.
• Then, how do you automate that without micro?  It would be something like a tree-traversing problem-seeker, which flags entire subtrees of your empire as all OK (green node) or this needs attention (flashing red node?).  Then your cognitive load collapses to linear in the problem deltas, which you hope are not proportional to the size of your empire, but maybe only to the size of your frontier.
• Each turn, you glance at the few red-hopping balls indicating a blown fuse or deadlocked queue, drill down as deep as you need to fix that, and you know you can skip the rest, because it's passed your checks-and-balances.  Surgical micro could be tolerable.  This is deliberately similar to unit testing: you want to know what didn't pass the tests.
• In SE5, I often have several actions running in parallel, all involving ships and planets, but for totally different goals:
• colonizers
• combat ships
• ships blockading enemy planets
• dome-lifting: You can colonize the wrong atmosphere, but you get a dome cutting it to 20% capacity.  Conquering planets gives you native breathers of those atmospheres.  Move some of them back in a transport, then take all of your non-breathers off-planet.  The dome vanishes, and you can build stuff.  Challenge: Do this without spacing any colonists out the airlock.
• SOSUS: Satellite-with-sensors (range 5-7), transport(s) full of them.  Laying a pattern of 6 grants full-time view of 1 entire sector.  This involves many small rendezvous problems, since any planet can build units (satellite is-a unit), if you just give the order enough in advance.
• remote mining: Asteroids have resources (just like planets).  You need medium satellites + robo-producers (one for each resource type).  Build those, transport carries them to every sector's 2 or 3 asteroid fields, launch them.  Gradually you can move vast portions of your production entirely off-world this way.  More rendezvous problems.
• trophy carrying: Conquered planets give you surviving enemy satellites, and (very rarely) weapon platforms.  I love to use a transport to haul these back to my homeworld, for the sole purpose of bragging about it.
• build to the helix: To conquer a homeworld, you (research and) build ground units, on any planet, at any time.  Any cargo ship can pick them up, but you generally need one small transport chock-full, and you'll take 30-60% losses.  Ergo, this is beyond-rendezvous: the ship traces a helix in space and time, and every planet along that route shall build up to the height of the helix at that point, and no more.
• thenardier: a boarding party + shipyard ship, to scavenge enemy ships I've crippled.  It meanders between furball hexes, picking up their pieces when they can't shoot straight.
• How do you manage all of those?  A single flat ship list (SE* provides that, too) is utterly lousy; just seeing 1 ship does not inform you what role you've grouped it into.
• After pondering, I envision a new primitive: provide customizable slices to each entity-list, where players can freely define and split off their own slices / subsets / groups.
• Let each slice mean one player-determined thing.  Then slices naturally handle regions, theaters, fleets, hierarchies-of-fleets, diplomatic gifts, garrisons, and any other concept that can be expressed as grouping.  This concept exists in other apps: layers in a graphics or CAD editor, groups in PowerPoint, themes in music, maybe even classes in C++.
• Each slice merges both entity list and map UI, e.g. you can see a slice as an overlay on the map.  (Whimsy: you can literally drag slices around "vertically" (in a 3rd dimension orthogonal to the map), e.g. to get them out of the way, or focus on them, or compare 2, or as part of the (gasp) cross-slice rendezvous stitching UI.  So if I want to restrict map view to only my attack-helices slice, then the UI suppresses all other slices.
• Slices can gain interaction flags (e.g. "needs attention"), pieces of code (satisfaction predicate, bean-counting, etc.), map icons, state flags (is-a-doodle, is-committed), etc.  Basically, once they become first-class UI elements, whole vistas open up.

Many of these ideas also cover war boardgames (with hex maps, c.f. Avalon Hill), probably for similar reasons.  We need UI tools to make what-if tests, compare alternatives between 2+ strategic choices, and generally see higher-level information superimposed upon (or floating above) the map -- which creates interesting problems in data visualization.

• Conquistador (boardgame) turn-planning consists of buying in advance the movement points for all of your ships in play.  To do that efficiently, you gotta doodle out all of their actual movements, right?  So what you really need to do is to plan your entire turn out, in breadboard mode.  Then you need a UI that provides doodling as the primitive.  So: slice-as-doodle, allowing you to make what-if slices at any time, and then finally commit to one.
• Most boardgames give you innumerable tactical battle problems, where you must decide how to allocate your factors to maximize odds.  What you're really choosing is between different configurations of good-odds.  In real boardgames, you physically tweezer the pieces around.  Slice-as-doodle works here, too: region-select a focus area (your units and all nearby enemy units), say "split a doodle-node", it floats above the map.  From that node, split a slice-as-doodle, make a configuration (here you might dip into click-hell, it's unavoidable), click Eval and see your odds.  Repeat this to split off many slices from that node; the UI organizes it as a tree (of depth 2).  Collapse all doodles into that node, expand them out, doodle more.  Repeat until you've spanned the entire tactical board.  Select one tiling of the board (a set of slices-as-doodles that cover the board, so that all game rules are satisfied, e.g. soak-off attacks at low odds to satisfy stack-adjacency rules), commit to the whole bunch.
• We do this in chess (in going-over-the-game mode, post-mortem) by physically moving the pieces around, then looking at the board, then moving them back ... or, OTB, you really do it all in your head, maintain the whole tree of alternatives, and evaluate them all.  But chess is a sport because it's so tiring, and we play games to have fun.  Good UI in FullHD is part of the experience.

I've focused almost entirely on the visualization part of a game UI, and kind of waved off the issues of diplomacy, game length, and immersion.  It's because a bad UI is such a roadblock that it eventually (or quickly) trumps everything else, and no amount of 4X fun can overcome it enough to keep me hooked.  Bad micro is  the antithesis of fun.  If I ever get around to writing my own, it'll have boring graphics, generic components ... but a script language capable of writing a C++ compiler (hehe), and UI elements that let you compose and doodle.  It's such an unexplored niche (even today).

As for the rest, I lack experience, so I'll abstain.  I've never seen a diplomacy worth beans (so I have zero expectations), game length depends on how inefficient the UI is (I almost never finish any 4X game), and nothing immerses me yet.  I find beauty in numbers, and in running an empire just a bit more efficiently than the other guy.  I'm open to any kind of game world machinations ... but nothing can mask a poor ability to juggle the numbers.  If they're important at all, then let me run an entire empire's worth of them my way, to my satisfaction, without demanding my constant attention.

i.e. let me defun

Greetings!

Today I would like to ask you guys some questions about strategy games.

1. What specific features of diplomacy do you traditionally like the most? I want you to be as specific as you can be. Which parts of diplomacy from any game do you like the most? What parts do you remember long after playing the most?

- I like Elemental : Legendary Heroes research trading system verses the common tech trading systems that other games have.

- Don't allow trading of credits. It turns it into a zero sum system that isn't that fun. Allow trading resources, tech credits, ships, planets ... etc. But do not allow credits. Also, do not display the perception value that Elemental displays. 

- Civ 5's BNW is pretty decent as far as diplo goes in a 4x game. 

2. Looking back, how many turns do your favorite games last? This is important to know the specific number of turns the game in Question lasted. 

- My last 5 games of LH is 208, 212, 264, 200, 232. I'd say around 200 on medium map. 

- Civ 5 games are about 350 turns ea. A side note on this. Between GnK and BNW the game has a number of systems that makes a 350 turn game not feel like forever. 

3. Consider all The 4X strategy games that you have ever played. How do you define what is a good strategy game or a bad one? To you what makes one strategy game good fand another one bad? Consider different memories you have of those games can you remember the parts that made you enjoy that game the most?

Here are some traits of really good games. Some of these are suggestions for LH 2.0, E2015, as well as GalCiv3

- Combat strategy is the most defining factor. AoW3, 

- Combined arms. Civ5

- Oh f**k moments. Such as XCOM. I like to see something that either is scary or let's me know I dun f'd up. Master of Magic and LH both meet this criteria where you have some really strong monsters near your starting area that you can't beat with your starting army. 

- Zone of Control is a must. I should be able to position my units in a way to slow the enemy. Civ5, LH tactical (can the strategy lair get something like this too)

- Flanking and for games not in space back attacks and positional attacks. 

- AI personality. Civ4 and Civ5

- Good economy. Civ5 BNW nailed the economy game I think. 

Traits of bad games

- Crappy music = crappy game. I've been put to sleep playing some 4x games. 

- Too simplistic combat

1. What specific features of diplomacy do you traditionally like the most? I want you to be as specific as you can be. Which parts of diplomacy from any game do you like the most? What parts do you remember long after playing the most?
in order of preference:
1: "declare war" is the one i like the most because it's the easiest way to expand once the initial rush for colonies has stopped.
2: "non aggression pact" is second, it means basically "i'm too weak to declare war on you so i need a bit time to prepare"
3: "trade treaty" and "technology treaty" which is just "gimme the means to prepare to declare war on you"
4: "make peace" because "my unrest is too high for now, we'll play again later"
5: "give gifts" because "i'm not ready to declare war on you, here is a toy, so leave me alone until i'm ready"
6: "trade things", useless because "i don't trade things with guys i intend to conquer except if you have things i need to prepare to declare war on you more quickly"
7: "declare war on this guy", useless because "no need, i'll take care of that myself and i don't want you to steal the things i want to conquer"
8: "trade technology", my least favorite one because usually, it shortens the game and i don't like the fact that the exchanged technology are immediately available to use on a large scale.

As to why, i'd say that's because i'm playing to win and have fun, and conquering and expanding is the easiest and most immediate way to achieve win and fun.
War is the easiest way because there is little drawback to going to war and a lot of rewards.
War is the most fun because strategy games focus a lot of their gameplay in the combat system.
Therefore, diplomacy is either a funny toy to play with or a tool to stall the AI until you're ready to go to war.

So, in order to stay in the subject, i'd say that the diplomacy i'd like is one which would bring as much in-game advantages than going to war AND an equally fun diplomacy system than the combat system.

About game examples, i'd say Star Wars Rebellion (or supremacy... wasn't there a naming issue in the US?) in which you sent diplomat characters to convince planets to join your side. A simple system but a fun one because you have diplomatic conquering units that were used almost the same as military units.

I also liked some parts of the Paradox games Europa Universalis and Hearts of Iron.
The first one because you could not conquer completely one contry in one war (except for small countries). When asking for peace, the winner got victory points according to occupied ennemy territories and he could spent these points to annex some of the territories of the defeated country.
There was also a drawback for going to war unprovoked (lose stability points, a very important variable)

And since trading technology is part of most of the diplomacy systems, i'd say the way Hearts of Iron handle the technology exchange through the use of blueprints is the best i've seen until now. Blueprints means that after acquiring technology through diplomacy, you just have a bonus to develop said technology, but you still have to develop it.
I always saw research points as an abstraction for the developpement process as well as for the industrialisation process. I mean, if tomorrow a research center develop quantic computers technology, it will still take years before you buy a quantic computer the same way you buy a nowaday computer.
But in most games, when the technology is available, it's immediatly operational and can be build in case of buildings or integrated in ship designs. It means that the prototyping, all tests, user feedbacks and standard production solution have been done/solved: it's more than just books and scientists knowing them, it includes the society assimilation of said technology.

One last thing about technology is the fact that Europa Universalis gave technology research bonuses to all countries less advanced than the leading country in a particular field: it represents the fact that ideas travels beyond the frontiers and that it's impossible to have 2 civilizations next to one another, one using smartphones, satellites and airplanes and the other using telegraph, steam engines and horses.

2. Looking back, how many turns do your favorite games last? This is important to know the specific number of turns the game in Question lasted.
The longest possible. Same reasons as other players.

3. Consider all The 4X strategy games that you have ever played. How do you define what is a good strategy game or a bad one? To you what makes one strategy game good fand another one bad? Consider different memories you have of those games can you remember the parts that made you enjoy that game the most?

Good points from a military point of view:
- Slow pace, i do like starcraft and the likes, but i prefer turn based for the impression to think strategically and not tactically.
- Big, limited and powerful military units such as heroes in fantasy games or powerful ships in space games (like admiral ships in Sins of a Solar Empires). It makes me really feel like the leader of a country with powerful and symbolic units. Real life examples are numerous, from the USS Enterprise to IJN Yamato, HMS Hood and many other.
A simple feature of Hearts of Iron was the fact that you could check your ships to know which ones it has sunk, if any. It brings nothing to the game, except once a ship has a glorious hunting score, the player will try to keep it alive, so it brings something to the player.
- The ability to retreat (not always available, i'm looking at you, Legendary Heroes!) as a meaningful strategy: the cost of healing/reparing must always be far less than the cost of rebuilding else, it's simplier to fight to the end which is not realistic.
- The ability to raise an army quite quickly: it's quite rare in turn based games and it's a shame because the ability to react quickly to a threat would make wars less easy for both players and AI.
- the limitation of force projection, Hearts of Iron tried a bit with its supply system but it wasn't this satisfactory the way they implemented it. Still, limiting the ability to send military units outside your borders make waging wars more difficult.
- the ability to design military units

Bad points from a military point of view:
- having to spend one move of an entire fleet to destroy only one ship on a stack of dozens of ships just because they're not in fleet.
- having no need of small units once you have big ones
- unbalanced weapons (like plasma cannons in revered MoO2)
- army building time superior to war time: it makes no sense that you can completely wipe out the enemy main army and then conquer most of its territory with little to none resistance since army raising takes such a long time.
- no retreat ability

Good points from an empire building point of view:
- Clear borders: it bugs me insanely to see unauthorized foreign military units cross my borders and make themselves at home. I was very happy with Civ5 about this.
- no artificial limits on expansion such as unrest, expansion limits should mimic reality (or what i understand of it ^^') such as economically inefficient colonies, population not growing faster than a horde of aphrodisiac-drugged rabbits and big infrastructures investments (roads, schools, health care, administration offices and such).
It's also okay to put an "administration expense" the more colonies/cities/population you have: "Bureaucracy expands to meet the needs of the expanding bureaucracy"
- Internal political choices, such as the social policies of Civ5
- Specialization of cities/colonies such as the fortress/towns/conclaves of Legendary Heroes
- good technology tree (but it's quite vague, i know x) )

Bad points from an empire building point of view:
- rush buy: a project needing 100 days for one man to complete cannot be done in one day by 100 men.
- new building/ship coming from a new tech that is less efficient that the building/ship of the old tech.
- no statistic/summary screens: i need to have lists of cities/colonies, of units, of armies/fleets, of buildings with their characteristics that i can sort and filter the way i like and from which i can jump to such or such selected item, be it a planet, unit or whatever. If you think you have put enough of these screens, you are wrong! There are never enough of these screens!
- unclear properties of a building or ability: i need to know what is what.

Well i'll stop here

For those who read all, well done!
For those who don't, don't worry, i wouldn't have read this wall of text myself x)

1.  I like when diplomacy isn't boiled down to a number.  Now, I know in the background that is always the case, but the games with the best diplomacy are the ones where the relationships between races look and feel like relationships, not an ever adjusting solution to an equation designed to calculate the relationship status.  If I've spent a lot of time and resources developing a bond with another empire, it should come across in the words used when speaking, the tone of the music when conducting diplomacy, the ability to request favors that other empires would almost certainly deny.  This all needs to be conveyed in as lifelike a manner as possible, as opposed to simply knowing that I have a +107 on the relationship scale, which I know makes us "blood brothers" because that's what anything over 100 conveys.  Diplomacy done right in a game is all about subtlety of the system and the appearance of intelligence in the CPU players.  This, more than any other, is the game system that can create the illusion of life in the CPU.  

1a.  Aside from that, I'd say flexibility in treaty options is my favorite actual feature.  "Let's work together to conquer X" or "Let's pool research to develop Y technology" or "Help me out by conquerering Planet Z to deprive my enemy of their critical resource A"  or what have you.  A layered system of short term and long term partnerships that really let you achieve serious goals through diplomacy.   

2.  As many as is feasible by the game's systems.  I play on the larger maps with lots of races.  I like busy, complicated games that last a long time.  In some games that's five hundred turns and in others it's thousands.  The one key here is that the game's systems, research in particular, often don't scale well to larger games which leads to hundreds and hundreds of turns of researching "Generic Tech X for +5% benefit to Y".  It works as an easy way to let the research system scale to game length, but otherwise it's pretty yawn inducing. 

3.  A good strategy game has to react to what the player is doing.  The environment and/or other players have to show that they care what the player is doing.  Not to pick on Elemental, but if I'm in the process of building the "Great Holy Towers of I'm About to Kick Your Ass", the other players should care.  My allies should be asking if I remember that we're allies. My enemies should be threatening death and destruction or they should be kissing-my-ass for a last minute alliance.  What they shouldn't do is not acknowledge it at all.  That's game ruining.  Conversely, the player must be forced to react to the environment and other players at times.  My allies should make the occasional onerous demand.  My enemies should sometimes catch me by surprise.  And amidst all this, things need to behave predictably.  I should be able to count on the fact that colonizing a very valuable planet near an ally, one that is maybe closer to them than to me, might upset said ally.  I should be able to count on cause and effect to be rational within the constraints of the players, factions and universes.  If a faction is unpredictable, the fact that they are unpredictable should be predictable.  Things need to work the way they are supposed to.  All of this combined is what gives a good strategy game the illusion of life and what makes replayability achievable. 

1. Alpha Centauri and Civ 4 were probably my favorite diplomacy. The mix between the diplomacy and your social engineering or religion presented a lot of interesting dilemmas - do you go with a particular social setting or agenda to court favor? Do you try to convert your neighbours to form a giant block of like interests? What I like about it is the way it inherently evolves into alliances and hatreds as factions pick sides, as well as the not entirely reliable input your actions would have on that process. Room for both triumphs and setbacks.

But I'll admit it was maybe a bit too easy to game. Still, not by much - I intensely dislike extreme - note, some is necessary! - unpredictability in the AI. A solid alliance ought to mean something - but hey, if an AI wants to break an alliance I'd at least like an explanation such as bribery or coercion, y'know?

2. Longer games, definitely - not in the thousand, but 300 or 500 turns I can certainly see. I only occassionally dip into shorter matches.

3. What makes a strategy game good? I'll echo 'interesting choices' from everybody else, and I'll also add the ability to create a kind of immersive narrative for yourself. I'll get back to interesting choices, but atmosphere is very important. AC's tech voice overs, music, the rumble of drone riots.. all tell a story without much need to actually have a story there. But there's also game mechanics - I still can tell the story of a multi-faction world war that erupted for me once in Civ 4, where my coalition was attacking another coalition through a neutral third country that was holding on to peace with each side - that story wouldn't've worked without alliances, without client nations, without their separate religions and ideologies, etc, etc.

Interesting choices - it's important to create 'apples vs. oranges' situations here. Where one option is inherently better than the others, that's not interesting at all. Choosing between a weapon that does six damage and one that does 8 is bland - but choosing between quick effective weapons and attempting to capture enemy ships with boarders? Ah, now that's interesting!

Moo 2 excelled in these kinds of combat options. In another genre, take a look at "FTL" and it's recent expansion the advanced edition. Or 'Mark of the Ninja' where you can kit your ninja out with smoke bombs or lethal traps, depending on how you want to play. Not just in combat, either, a 4x must be about interesting apples vs oranges choices -everywhere-.

 For a bad strategy game example.. oooh. I don't want to diss it because Endless Space is awesome in so many ways. But the combat system! Talk about the epitome of 'I don't know what impact my choices going into this battle are actually going to have'. It's a game where combat between ships has several rounds, and you can pick a tactic that theoretically has benefits depending on what your opponent does - but the computer seemed to make it's own choices randomly, so getting the intended effect you wanted was trying to guess the computer's move in advance - completely unpredictable, unsatisfactory.

That's also a game with some bat 'apples vs apples' choices. Sure, I could make my lasers focused for short range combat.. but since they'd be inherently weaker on anything other than the long range combat they were best at, what was the point? Terrible design.

I see all 3 of these questions as inter related.  I will play a game for hours if I feel like the universe is dynamic and diplomacy makes sense, meaning that my ally does not suddenly and for no reason attack me.  As stupid as Britain suddenly turning on the US during WWII.  

I would like the option to work together with a ally to develop a tech like the US and GB did with nuclear weapons.  Share intelligence and conduct joint invasions, or liberate planets that were previously under an ally's control.

The best games I have played immersed me in the game.  I feel like the wold/ galaxy is reacting to my moves.  If I am a ruthless conqueror the other factions join together to check my progress or try to jump on the band wagon .  I like to create custom ships that I care about and even better my faction cares about.  Think the impact the USS Enterprise has on Starfleet as a symbol of discovery and defense.  Its loss has a impact on everyone to the point that they recommission ship after ship with its name.  Included with that, the loss of a fleet should be a big deal, not just fodder.  That should have a impact on moral.  Both for my people and my enemies.  A crippling loss should bring people to the peace table.  I hate when I essentially have to completely destroy another faction to end a war.  Its again very unrealistic.  

I have also liked games that gave me a sense of history being made in the universe I was playing in, giving names to wars.  Chronological events in a dramatic narrative.  Little things that give the impression that may actions make a difference is always nice.  I don't want to waste time just repeatedly clicking next turn hoping for something to eventually happen.  Finally, I love games that make peace fun.  Every game, more or less, is fun when your at war.  Battles and ships ect.. But achievements of engineering, science, and economics.  That takes skill.  Say construction of a gate travel system or a Dyson sphere would be amazing.  The first major star base or something would make the peace fun too.