In answer to this question. I am not a "believer". My initial position is one of more trust in the scientific process of peer review and debate than you. It is not that individual scientists are great - but when many different people are all contributing different ideas, the best win over time. And GCMs have had time.
When I look at the evidence now my problem is that this is complex science. The simple-minded blog rebuttals do not stand scrutiny. Equally, I can't rule out completely that all the pro-AGW scientists are somehow deluded / in collusion and getting it wrong.
Clearly the data for warming/cooling is not simple to analyse (take for example the sea-level data, or the horrible surface station data). Few of the sensors were designed to detect fractional degree global temperature change and the one recent satellite that would have helped failed. The anecdotal evidence (do we in the NH think the last few years have been hot or cold) is essentially worthless, partly because it is global temperature which is the issue, and which has less medium term noise than regional temperature.
If you look at temperature since 1979 you can see there is a lot of noise. The current temp is at the low end of the moving average - and if it continued with no increase for another 5 years you might doubt the trend. As it stands this data is not (I guess) statistically very significant for a trend though it seems likely there is an upward trend.
So if I had to work out trend based on just this data I would not feel it was sure. The GCMs do not work out trends based on this data. The use all sorts of other data to determine parameters. And my understanding is that they do not determine any parameters from contemporary temperature record. Looking at the noise in this signal it would be difficult to do so.
Similarly - another 5 years will make me more or less happy with 0.2 degrees/decade as a linear trend. But probably this data would not settle it either way.
However the models are getting much more sophisticated. You would expect that they can get some of this fluctuation from independently measurable parameters, ocean currents etc. If they do this well any underlying trend becomes clearer.
I don't know how well they are doing this. Nor do you. Blog comments saying that they are doing it badly for some simple reason are not credible (none of these simple reasons that I have looked at survive close scrutiny). But equally I can't prove that they are doing it well, and not modelling noise, without going into the GCMs in as much detail as the climate scientists.
Now, here are some figures, taken from realclimate (biassed you may think, so we will discuss issues later) comparing the dreaded Hansen's 1988 model projections with global temperature. This comparison is not simple - and anyone who makes it wthout thinking carefully about conditions will make mistakes.
First, all these projected models have as INPUTS the overall climate forcings in the future from GHGs. This is got from the relatively uncontentious proportions of these GHGs in the stmosphere and their raw affect in climate via absorbtion spectra. No assumptions here about H2O amplification, that comes from the models. Hansen proposed three scenarios in 1988 and the one which is more or less correct is B.
Now look at as OUTPUT the projected global temperatures for these three scenarios and the measured global temperature since then:
This was produced a few years ago, and misses the last few (outlying in the cold direction) temperature points. So the pretty good fit over this time period will look less impressive if you add these points.
More recent models are better than these old Hansen projections because:
(1) they model oceans and ocean/atmosphere interaction more precisely
(2) they model vegetation, ice feedbacks more precisely
(3) they have more computer power available, can do ensemble projections with slightly different initial conditions (Simon's Monte Carlo).
Modelling solar input as a forcing is problematic because there is no agreement yet as to what is the overall affect of more/less solar wind. The papers i have seen make this affect a relatively small forcing however.
I am not certain about how the latest models deal with aerosols. As you know there is indisputable evidence that particles in the atmosphere tend to decrease temperature. it is complex - some particles at some heights can have the opposite affect by decreasing albedo - but overall volcanic eruptions make things colder as did industrial emmissions before the major air quality increases of the 1970s. As always take one little bit of this complexity and write it up on a blog and you can argue for the reverse effect!
Recently industrial emmissions from China and India have been rising at a very rapid rate and maybe they are now significant - sorry I will need to do some research to know whether or not this is an issue, and also how it is incorporated into current models. But it was not in Hansen's.
I realise this in no way settles the matter. It leaves me without evidence to change my initial position against the consensus. But it does not either indisputably back the AGW case. You need a lot more detailed consideration of how the projections perform to do that.
The recent deviation towards cold (the last few data points) is interesting. It will be a once in thirty years or more deviation if it continues a few more years. It probably is explainable (if not just a random fluke) it will be interesting to see the explanations. Maybe they will have a significant effect on my view of the overall trend and its cause.