Another thing: why don’t we have proper Greek typography? Why is Greek always printed in fancy but not so readable italic sans-serif fonts? Why don’t we have the equivalent of easily readable fonts like Times for Greek? Reading even your own language in italic tires you quickly and makes you slower - but that sort of font is about the only thing we have for Greek. The most readable Greek font I’ve seen is the one used by Budé, and even that is sans-serif, like our Helvetica. Have you ever seen a “real” English book printed in Helvetica? And this isn’t just a question of getting used to it, it’s for a reason the Germans gave up their fraktur and all the newspaper fonts look the same nowadays all over the Western world.
I did a quick google and came up with this http://www.uv.es/~mperea/serif_JCP.pdf which suggest that the evidence that serifs help is sparse and inconclusive. It may however be that, when reading a language that is not your native language, serifs are helpful.
Not sure I can keep up with this flurry of activity. 
First, apologies to Scribo for misunderstanding.
Thanks to Paul for the WSAllen quote. Vox Graeca remains the Bible on this stuff.
I don’t think there can be much doubt about compound verbs (exanesthke e.g.) continuing to be felt as compounds. How could they not be? And e.g. afesthke confirms the aspiration (assuming the representation to be trustworthy, but I don’t think it’s ever queried, except perhaps in some post by Qimmik for tragic dialogue). But I expect there may in practice have been degrees of aspiration according to dialect and kind of word and individual, so it may not always be a black&white either/or matter as far as the actual phonetics were concerned (at least in non-psilotic dialects, but perhaps even there). But etymology tells us what “should” be aspirated and what not, and phonology seems generally to conform. And there’s no difference in principle between word-initial and word-internal aspiration. (Quite apart from the What-is-a-word? problem.)
Int, I’m afraid any Orthographical Spring is going to have to come from the top not the bottom. If some respected and powerful editor were to insist on printing texts without smoothies and with internal roughies … Martin West where are you in our hour of need?
I’ll leave modern typography for another day. I rather like the OCT font myself, and dislike the Bude. Not sure why, never thought about it. I find the old OCT Plato looks far better on the page than the new one, mind. – But don’t let’s switch the thread from orthography.
But what makes the internal aspiration disappear then once the word “was no longer felt as a compound”?
I know there was the phenomenon of intervocalic sigmas disappearing via /h/, but I think that happened too early. And it wouldn’t explain phil(h)ippos.
Typography: I guess you’re right, Daivid, that serif or sans-serif is not the point. But I think italic/oblique fonts are clearly more difficult to read. Of course, it depends on the font. But at least I find it annoying to read longer sections of text in italic, even in my own language.
Mark that a readable font isn’t necessarily a very beautiful one and vice versa. The Budés are all extremely readable, though I prefer their older font (or one of the older ones, maybe there are several?) The old OCT Plato is nice but it’s taxing to read (if we’re talking about the same one).
Look at this Hesiod from 1772 I found at an old books store in Lissabon in September, a find for 20 euros. The Greek font is very beautiful, but it certainly illustrates the tradition of frivolous Greek typography we still haven’t got rid of. Sure Latin is dull compared to Greek, but you don’t have to show it with typography!
Ah two of my least favourite things: Macs and old text editions.
Ahhh! - it’s unreadable (and to me ugly). But be fair, most modern texts have nothing close to that.
Well, /h/ is easily effaced, isn’t it (and cf. Cat. 84 for hypercorrection)? It’s a very volatile sound. But a proper name such as Filippos, which does indeed seem to have lost its internal aspirate, is very much the exception rather than the rule, and we shouldn’t allow it to sidetrack us. (But we are all so easily sidetracked, aren’t we?) In the case of ordinary compound verbs, adjectives, nouns, whatever, there’s no reason that I know of to imagine that the aspirate disappeared.
Of course it’s difficult to track the phonetic realities. Educated people wrote as they were taught, not necessarily as they spoke. Inscriptions provide a certain amount of evidence, fairly hard evidence when they deviate from “correct” orthographies. So do puns and the like in Aristophanes and elsewhere. And so do papyrus texts written by semi-literates. Without conducting an proper survey I have the impression that they indicate no more loss or vacillation of word-internal aspiration than of word-initial.
So the question remains, Why don’t we mark internal aspiration? Is there any less irrational answer than “Because we don’t”?
Maybe you should have another poll on this, but given the resistance your new poll shows to the idea of dropping the absolutely functionless smooth breathing, I’m guessing it wouldn’t get much support.
I agree, much too early to be relevant.