Like most of you (I gather), I reread "A Song of Ice and Fire" about once per year, or near enough as makes no matter. For this reread, I want to focus on the structure of "A Game of Thrones", a kind of meta-analysis, if you will, and especially concentrate on "early installment weirdness". That term relates to the first volume of a series, or the first episode, or whatever your medium might be, and how it usually is a bit rough on the edges. Ideas are not fully formed yet, characters not really "there", and there are elements that didn't work and were consequently dropped.
There's a surprisingly big amount of that. It shouldn't really be surprising, given that Martin started writing this almost thirty years ago (in 1993!), but compared to the later novels or the (finished!) TV series, it is rather noticable. For example, there is Tyrion's artistic ability, never brought up again until Martin retroactively put a lid on it in "A Dance with Dragons" by explaining it as an artifact of his backstory.
Such details pale towards plot elements that stem from Martin's original outline. There is the groundwork laid for the later three-way-romance between Tyrion, Jon and Arya (mercifully dropped already by the end of the novel, but if you know that Martin planned for this, it's clear to see). Jon's relationship with Arya is strongly established and has little payoff in the novels following it; Robb is the much more important fixpoint for Jon's memories of home.
There's the groundwork laid for the later planned destruction of Winterfell at Tyrion's hands. The wolves' aggression towards Tyrion, his strong connection to the place, the tragic of his friendship to Jon - both in terms of the planned romance with Arya and the destruction of his childhood home - all point towards that direction.
There's the groundwork laid for Catelyn's journey beyond the Wall, as when she is the primary conduit for the dark premonitions about the Land Beyond the Wall, Mance Rayder and the Others. When she hopes that Eddard will have gotten her pregnant again after they had sex in her first chapter, we see an echo of the child birth that was supposed to be her death in that frozen wasteland (a plot thread that Martin returned to with alarming regularity since).
There are also many elements that, would he write "A Game of Thrones" now, would be there but are absent. The most glaring for me is the lack of references. Eddard Stark becomes Hand of the King, but no one compares him to Cregan Stark, which would be an obvious comparison, especially for Pycelle, Varys and Littlefinger. The behavior of people towards Eddard as the first Stark Hand since Cregan makes no sense at all now that "The Hour of the Wolf" is a thing, but of course, it was not yet conceived back then.
People instead tread Ned as a provincial, a bit unrefined and straightforward, much as he is written. But given what we know about Targaryen history by now, there should be a lot darker and much more concrete biases at work. The same goes for kingsguard, king and nobility in general, the Dothraki and the role of the Free Cities - none of it is grounded in the detailed history Martin has written since. One can debate, I guess, the wisdom of creating all that stuff afterwards.
The same is true of several regions: the Iron Islands are treated as an afterthought; they will be developed as a solution after Tyrion cannot destroy Winterfell anymore because his plot leads to King's Landing. Dorne is only mentioned in passing. The Tyrells likewise. There's no mention of the Crownlands. And so on.
The novel itself remains the weakest of the entire series when we talk about intricacy of plotting and depth of character. It is "only" an extremely well written political thriller set in a low-fantasy world. The main threads are Ned's investigation in Jon Arryn's death - a mystery that will only be solved in the finale of "A Storm of Swords"! - and the political fallout of the Lannister intrigue against Robert Baratheon.
What is very noticable is the tight plotting on the one hand - chapters are following directly on each other and deliver the consequences of the actions of the previous chapter much of the time, instead of following unconnected threads, with the notable exception of Daenerys' arc. But even Dany gets connected to the main plot via the murder attempt and the fears of Robert in a way that will not be true in the following novels.
It's even more pronounced with Jon's arc, which is so carefully plotted that each revelation comes just too late for Jon to take a different course, perfectly calibrated to play out his inner struggles with his dual identity between Stark and Night's Watch.
That is not to take away from an, once again, extremely well written novel. But especially compared to Feastdance, the lack of themes, the close interconnectedness of character arcs with the plot, and above all, the careful construction of the plot stick out. It is incredible on how many chances and coincidences the plot hinges. The fates in the person of Martin have their thumbs on the scales, HARD.
Once again, all of that is not take away from "A Game of Thrones". It makes it, however, the least "A Song of Ice and Fire"-y of all the novels. It's no wonder that Martin was able to write the first three novels so much faster than the last three. The main challenge here is to think about which character best to tell which event through, as to obfuscate and set up most effectively. But there is no question who is present where when; Martin only needs to choose. There is no Meereenese Knot, no question of which character will arrive when where to which effect, how to make time jumps and so on. It's almost quaint. And if you know "A Game of Thrones", you know what that means for the series at large. It's a breathtaking accomplishment. One can only stand in awe of Martin's abilities.
