First, this game isn't exactly what you may have heard. Don't expect a real time version of Gal Civ 2, or a sort of Europa Universalis in space, this is an RTS game first. Game play consists of collecting resources, building ships, and hurling big piles of them at big piles of enemy ships. It has some characteristics of TBS space games, like warp lanes, planet hopping, and a broadly slower pace. It also has some light approximations of the research and infrastructure elements of TBS titles, but there's no empire-building here to speak of. There are also no victory conditions beyond "annihilate the other guy." The bulk of your time will definitely be spent on traditional RTS activities, so if you don't care for that genre, don't invest.
I have played this unique title for several weeks now, and come to some conclusions. It's an interesting game with a steep learning curve for an RTS. I finally have a good feel for the rather unique interface, and it works reasonably well. The empire tree is a novel tool to control building and to try to track structures and ships in your empire. It lacks a sense of relative position, however, and as such I still find myself missing a mini-map. The main problems, however, all stem from one simple component of the design: the real time battles take place in the same timeline as the real time strategy. That is, while your ships are fighting, time is ticking by all across your empire. Got 3 battles going on? You can only watch and manage one of them. The AI's ok at handling fights for you, but I dislike being reliant on that. For that matter, the graphics are gorgeous and the battles are genuinely exciting, it kind of sucks that at best I can only watch one at a time and frequently I'm pulled away from that to handle managerial issues. The empire tree mitigates this at times, but fails to do so at others. It's not the universal control panel that I believe the designers intended it to be.
It's probably also worth mentioning that there are only three factions, and they play quite similarly. The art and voice work for each is well done and quite distinct, but the vaunted unique tech trees are actually about 75% equivalent to one another, at a guess. The result is that there's little different between factions beyond the aesthetic.
The single player diplomatic system is peculiar. I'm not fond of it, and it bears a few words simply because I've never seen a system like this before. In a free for all, single player game your opponents will give you "missions" to curry their favor. These can be straight-up extortion, like "give us X amount of this resource" or they can be something like "kill X number of ships/structures belonging to this enemy." Performing the task in the time allotted will increase relations with the faction in question, failing to do so will lower relations. What's a bit strange is that even if you're actively engaged in hostilities with an opponent (like, say, bombarding and recolonizing their worlds) they'll still message you with these missions. Even when you've made friends with a given AI, in many cases performing these tasks are a practical impossibility. Imagine, for example, being in the midst of a massive war with one enemy when your "ally" demands that you destroy a large number of ships and structure from a different enemy in a short amount of time. This kind of thing causes you to lose allies when you need them most quite often and for no good reason. If it were a bit more context-sensitive (ie when you're fighting a joint war, the requests involve the destruction of the opponent you're both actively attacking), it'd work better. Since your other actions in the game beyond these "missions" affect your relations with the AI players not at all, as best I can tell, I find it's best to simply play with locked teams and avoid the system altogether. The result, though, is that the diplomacy element of a 4X game is either broken or missing, here, depending on how you play.
On the positive side, the random map generator is the single most powerful function of its type I've ever seen in a game, bar none. It's better than the map generation in the Civilization series, and that's really saying something. There appear to be hard caps on almost nothing, you can create a map of any size to virtually any specifications you can dream of (and that your PC can handle), it more than compensates to the lack of a story-based campaign IMO.
It's also worth noting that the game is very modable. With a sufficiently engaged, motivated fan community (which this game appears to have, at this early date) and with the level of support and additions I expect from a Stardock product, there's reason to think it'll only get better with time.
So for what it is, I'd say the game is pretty well done. It seems reasonably polished, very stable, the AI's adequate while you're learning, and the graphics and sounds are great. It scales well to low and high end systems. The capital ship level up system is neat (if borrowed completely from Warcraft 3) and the battles are quite cool looking. The problem, at heart, is that the large scale strategy (which already has a few flaws, as mentioned) and the tactical battles conflict with, rather than complement, one another. It's two games that don't quite fit together, and thus make one flawed game when all is said and done.
But don't take my word for it. The best way to find out about Sins is to play the demo, which the devs have said will be out within a month or so of Sins' release (that is, sometime in March '08). You are wise to wait and do this, as there's never been a game quite like this one and whether or not it works for you is going to depend a great deal on your personal tastes.