6. THE PATHOLOGIES OF COSMOLOGY

1. Cosmology must be the slowest moving branch of science. The number of practitioners per relevant observation is ridiculous. Consequently the same old things have to be said by the same old people (and by new ones) over and over and over again. For instance ``Cold Dark Matter'' now sounds to me like a religious liturgy which its adherents chant like a mantra in the mindless hope that it will spring into existence. Much of cosmology is unhealthily self-referencing and it seems to an outsider like myself that cosmological fashions and reputations are made more by acclamation than by genuine scientific debate.

2. There is a serious problem with the cost of astronomical spacecraft. An instrument capable of cosmologically interesting observations may cost half a billion dollars or more. There is therefore an insidious temptation to overclaim what they will see [1]. This, however, is a dangerous game which can blow up in your face, as proponents of the Supercollider were to find out.

3. There is something beguiling and yet fallacious about working on ``the faintest objects ever observed'' even though, by definition, they contain ``the least information ever detected''. During my working life a major fraction of the prime time on all large telescopes has been devoted to the study of objects right at the horizon, with, or so it seems to me, very little result. To be rude about it, statistical studies of faint objects can keep a career going for ages without the need for a single original thought - or indeed a genuinely clear result. The jam is always just around the next corner.

4. As particle physics has become paralyzed by its escalating cost many particle theorists have `moved over' into cosmology, wishfully thinking of the Universe as `The great Accelerator in the Sky'. Alas they are mostly not equipped with the astronomical background to appreciate how `soft' an observational, as opposed to an experimental science, has to be. But they have only to look at the history of astronomy and at some of the howlers we have made (Table 4) to find out.

   Table 4. SOME HISTORICAL MISTAKES IN COSMOLOGY

   1.	`Early' cosmologies - e.g. Genesis, Hindu, . . .
   2.	Many unsound explanations for dark sky (up to 1960).
   3.	Assumption of a static Universe.
   4.	Original expansion claim based on unsound statistics (Hubble).
   5.	H0 wrong by factor ~ 10 for 25 years.
   6.	Universe measured to be younger than stars.
   7.	CBR not recognised for 25 years [McKellar 1942, Gamov . . .
   8.	Radio-source counts misinterpreted due to use of fallacious statistics.
   9.	Mass of neutrinos forgotten/ignored for 40 years.
   10.	Sandage's ``search for 2 numbers'' forgot evolution.
   11.	Horizon/flatness problems virtually ignored before a possible solution appeared.

5. Despite our intuitions very many Inverse Problems (and astronomy is very largely an Inverse Problem) are not well posed. [10]. For example when the HST was found to be spherically aberrated half the astronomical community claimed that the images could be restored by mathematical `deconvolution'. But they could not be - because the problem is ill posed; the highest resolution information will be swamped by the highest frequency noise during the inversion - it is a fundamental property of numerical differentiation. Only very high signal-to-noise data (a luxury astronomers rarely enjoy) can be deconvolved successfully. Likewise, I suspect that the multiparticle simulations beloved of certain numerical cosmologists are extremely ill-posed. They start off with a whole lot of CDM `dots', the dots apparently form filaments under the force of gravity - as they are bound to do according to Zeldovich's simple back-of-the-envelope analysis, and we are supposed to admire the result. What result? That to me is the question. Presumably we are supposed to compare the dots with real structures and infer some properties of the physical Universe. In my opinion it is nothing more than a seductive but futile computer game. What about the gas-dynamics, the initial conditions, the star-formation physics, evolution, dust, biasing, a proper correlation statistic, the feedback between radiation and matter . . . ? Without a good stab at all these effects `dotty cosmology' is no more relevant to real cosmology than the computer game `Life' is to evolutionary biology.

6. However, the most unhealthy aspect of cosmology is its unspoken parallel with religion. Both deal with big but probably unanswerable questions. The rapt audience, the media exposure, the big book-sale, tempt priests and rogues, as well as the gullible, like no other subject in science. For that reason alone other scientists simply must treat the pretensions of cosmology, and of professional cosmologists, with heightened scepticism, as I am attempting to do here.