stringph raises the issue of cognitive dissonance - rejecting, ignoring, or explaining away evidence that does not conform to one's most favored theory. I concur that it is very important to avoid this. You will find words to this effect on the last slide of my review astro-ph/0606351, which enumerates some big problems for MOND as well as some of its successes.
I cannot speak for others, and would not wish to over-generalize about "MOND-ians" and "LCDM-ians." Certainly there are no lack of individuals who fall into the cognitive dissonance trap. There are also some who don't, and you will find sane voices on both "sides" of this debate.
For myself, it depends on how I weight the data. Crudely speaking, there is a scale (roughly that of groups of galaxies) above which the universe looks like LCDM and below which it looks more like MOND. Even so, both theories face challenges in the regime where they generally work "better." To give specific examples, on galaxy scales the velocity dispersions of the ultrafaint dwarfs are uncomfortably large for MOND, while on large scales voids are too large and too empty for LCDM.
To reply to the specific point about the hand waving in Swaters et al (which includes myself): first, it is absurd to accuse Rob Swaters of favoring MOND. Anyone even passingly familiar with him or his work would know this is silly. Moreover, anyone passingly familiar with astronomical data will be aware that they are often plagued by systematic uncertainties that render the usual statistical tests nearly useless. The data in this particular paper conform to my experience with galaxy data: as the quality of the data improve, so does their agreement with MOND. Conversely, the failure rate increases as the data quality decreases. Any practicing observer understands this.
The statistical tests have been done. In general, fitting functions for rotation curves, in order of increasing chi^2, go (1) pseudo-isothermal halos, (2) MOND, (3) NFW. Sadly, the best fitting pseudo-isothermal halos have no basis in theory and seem to work simply because it is a very flexible fitting function. NFW is the nominal prediction of LCDM, and works systematically worse than MOND in spite of having more free parameters. There are of course individual examples that do not follow this trend, but statistically this is what the data do.
As for the solar system, I think Blanchet & Novak (1010.1349) have made an important contribution in showing what MONDian interpolation functions are already excluded. I'm not sure those authors would concur with your characterization of their work, but you'd have to ask them. [I did discuss it with Blanchet personally.] It is not surprising that the excluded interpolation functions are the ones that would be most readily detected in solar system data. Otherwise we might already know about it from such data. There is no tuning here; the scale on which MOND phenomena are obvious is eight orders of magnitude removed from that of the inner solar system. Two interpolation functions that are hard to distinguish at galactic scales are x/(1+x) and 1-exp(-x) where x = a/a0. One predicts the inner solar system would see MONDian effects of one part in 10^8 - subtle, but readily detected in precision data. The other predicts effects of one part in exp(10^8) - completely undetectable.
If we are guilty of partisanship, then it is tempting to attribute the Pioneer anomaly to MOND, as it is the right order of magnitude. However, I personally have even been persistently skeptical of attributing the Pioneer anomaly to MOND for the same reason: even the Pioneers are still four decades above the MOND regime. If the Pioneer anomaly is really a MOND effect, it is huge. (Translation: God is shouting at us to pay attention.)
That is not to say it is hopeless to test MOND with precision data. Bekenstein & Magueijo (astro-ph/0602266) have predicted that MONDian effects may be perceptible at the saddle points of the potential between the earth and sun. If instead of a modification of gravity MOND is a modification of inertia, it may be testable in terrestrial experiments (gr-qc/0612159).
So yes, stringph, there is Science. There is also no lack of both reality denial and intense hand waving. The challenge to each of us is to chart our own course through the noise and the data as objectively as may be within our power, even if that means giving up a few cherished beliefs along the way.