Of course, earlier GPs are better, much better. We should try to find this old article, as I'd really like to know how much more GP you can really get. I was under the impression that it was not much more than 1 or 2 for the philosphical trait. My memory sometimes plays tricks on me.
No matter what, philosophical is still a very good trait, because earlier GP is definitely a great advantage.
Yeah, I'd also like to try a industrious/philosophical leader. With the wet dream of warmongers (Boudica (Agg/Cha)) now in the game, I don't see why builders couldn't have their dream leader as well...
If we assume that Phi does nothing but double your GP points (which is slightly biased in favor of Phi considering the Parthenon, National Epic and Pacifism but easy to understand), the effect is basically the same as if you'd get the GP for half the price.
Therefore, since each GP costs 100 more than the previous (i.e. a fixed difference), we can handle Phi as if it'd only cost 50 more. So under those simple assumptions you do get twice as many GP than normally (unless I'm doing some strange mistake).
If we didn't have a fixed difference but one that depends on how many GP you already had, you wouldn't get many more but since it's a fixed width you do.
Edit: Just got me a pen and paper and calced it through. Assuming n is the number of GP points divided by 100 (since we get a GP at every 100), the question is how many GP did we get in total at n?
Mathematically, we define this as an inequation:
sup (sum i from 1 to k) <= n, i.e. the greatest number k for which 1+2+3+...+(k-1)+k<=n or each GP weighted with their cost is still smaller than our current pool.
Gauß found out a few hundred years ago that the sum equals
k/2 * (k+1)=1/2(k^2+k) <= n
k^2 + k - 2n <= 0
Using the p-q-formula:
k <= -1/2 + sqrt(1/4 + 2n) (second solution omitted because it'd become negative)
If we take the maximum of all whole numbers that satisfy that inequality, we arrive at our total number of GP, k_max
For example plugging in 10 (or 1000 points) we get out k=4, which fits (100+200+300+400=1000)
Under the above assumptions, with philosophical we replace n->2n and get
k2 <= -1/2 + sqrt(1/4 + 4n)
So you were indeed right, we don't get twice as many. Instead, we get roughly sqrt(2) times as many GP than normal (for slightly larger n), so roughly 40%
As mentioned things get worse if we decide on some other function than n->2n, for example if we usually have Pacifism anyways we go for n->1.5n and respectively we get about sqrt(3/2) as many, which is only about 22% more.
However you do get the first one in roughly half the time and we all know that they're most useful at the start...
Still I don't find Phi that strong (although building the Uni cheaper is nice, too).
In case you wonder why my initial thought was wrong: Well it wasn't, I just didn't see that halving the number of points required wouldn't mean that you'd get twice as many GP because yes, you would get them at 50, 100, 150, 200, 250, 300 but by then you'll have a total cost of 1050 whereas with the normal rate you'd get them at 100, 200, 300, 400 where you have a cost of 1000. It just doesn't halve your total points but only the individual points required for each step.
more than the farmed tiles which is the same bonus you get from your specialist with representation. In addition to that, you only need two citizens to work the tiles instead of three.
than a CE. I think it was like around 400 
/turn. It'll take us 20 turns without Phi and 10 with - this is because there are no other multipliers. Say we build National Epic - the 
