I agree that in order to advance the debate on cosmology we need to guard against falling in the trap of considering the opposite views in the debate as overly simplified caricatures. In particular, the notion that scientific cosmology has primacy over other forms of inquiry into the universe just because it uses "facts" to discard false theories. The second oversimplification referred to by Ludwik is the assertion that constructivists consider the world as a product of the images of it as seen by humans. Implied in these mischaracterizations is that facts and values are separate entities. As a corollary, it then follows that every experiment needs an interpreter, hence scientific knowledge is a social construct. These are important elements that need to be included in this debate, but I think the analysis is incomplete. There are two missing caricatures that we must also strive to guard against, namely: 1. a theory is either real OR it is a social construct but not both. As if labeling something a "social construct" diminishes its reliability or validity. It is common among scientists to feel attacked when told that their theories are just social constructs. This is not an unjustified reaction given the implications (i.e. religion, myth making, and belief in the paranormal are also social constructs, so science is lumped together with these other activities). And, 2. "All social constructs are equally reliable or useful". This is a common perception that is unwarranted.
These two additions to the list of caricatures emerge in large part from shortcomings of the views put forward by sociologists of science. Social analysis of science has, thus far, failed in at least two counts: 1. it does not recognize that not all social constructs are created equal; and 2. it mischaracterizes science because it only looks at ‘projections’ of this human endeavor on the different axes they consider but not as a whole.
Let’s begin with the problem of gradations among social constructs. Yes, science is a social construct, and yes "facts" and values are not independent entities, but that alone does not imply that all social constructs are equally reliable. To explain what is meant by "not all social constructs are equally reliable" it is important to accept two premises: 1. there is an external world out there independent of it being observed; and 2. we can interact with the external world. In science that interaction is carried out by means of measurements. In addition to allowing measurements, the scientific activity also endeavors to explain those measurements. A measurement is just the straightforward comparison of an observed quality of the external world with a reference pattern, "the length of this table is two meters". Measurements are statements that we all can agree have the same meaning. Like language, once we agree on the definition of "meter" then two independent groups can have a meaningful exchange when referring to the results of measuring the length of the table. I posit that measurements are indisputable realizations about the external world and that theories are interpretations of these measurements and thus they are pronouncements about the external world that are social constructs. Modern cosmology offers a particularly relevant example to illustrate these points. The redshifts of remote galaxies have been observed by astronomers on both camps of the cosmology debate (i.e. steady state and big bang). The observation that remote galaxies exhibit a red shift in their spectra is a straightforward measurement that no one disputes. It is the interpretation of that measurement where the "social construct" comes in. For steady state cosmologists, the red shift has nothing to do with distance but it is instead a manifestation of galactic evolution, whereas for big bang cosmologists red shift is a distance indicator. I can easily be criticized by the simplistic notion that measurements are straightforward interactions with the external world. In today’s scientific laboratories measurements are not the direct comparison of the length of an object with a meter tape, they are the result of indirect interactions at times quite difficult to disentangle without the help of a theory. Example: the cosmic background radiation observed by COBE and WMAP is electromagnetic noise that after very tedious and complex analysis is disentangled between a component of noise due to the sensors and a second component interpreted as originated from cosmological sources. The point is that -- no matter how complex -- before the help of a theory must be invoked, it is always possible to reduce the measurement to direct observations of something.
Accepting that there exists the possibility to interrogate nature via measurement, and likewise accepting the neutral character of measurement, provides an efficient criteria to rank social constructs in a reliability and usefulness scale. Namely, those social constructs that are backed by a large and coherent set of measurements can claim a higher rank. Table manners are a social construct, creation myths are social constructs, astrology, voodoo and shaman healing are social constructs, and science is also social construct. But not all these social constructs can claim the same level of reliability and usefulness because their contact with the external world (via measurement) has different levels of development. This concept shields the "social construct" notion advanced by sociologists of science from the caricature that science can be lumped together with table manners and "quantum healing".
I believe that science studies have made an important contribution in ridding us from scientism, but at the same time they have served to support the irresponsible view point whereby any patently bogus claim coming from snake oil dealers or global warming deniers paid by the oil industry should be given the same footing just because they are all social constructs. I think that, in this age of global perils, it is quite irresponsible and unethical to proclaim in pompous and academic jargon that the utterances of a shaman have the same value as a scientific theory that has accumulated 100 years of consistent experimental results (measurements) and logical coherence. With that, I am not saying that one must accept science without questioning and criticism. Assume for a moment the hypothetical situation where you are responsible for administering the research budget of a small developing country. You need to decide between funding a proposal to study Cepheid variable stars or a proposal to predict volcanic eruptions based on astrology. Both proposals have the same cost, but you have budget for only one. Which one would you choose? You know that no matter what decision you make you will be furiously criticized by the looser team. Upon funding denial, astrologers will accuse you of being culturally bias and discriminating because you deem scientific teams superior than astrologers, they will argue that the study of Cepheid variables has no relevance for the people and the needs of the country, they will insist that astronomy is a luxury that only developed countries can afford, they will support their proposal by saying that they can save lives by predicting volcanic eruptions and that you are denying the possibility for saving lives. Scientists, on the other hand, will point to the fact that there is no precedent or any kind of backing showing that astrology can predict volcanic activity, they will accuse you of irresponsibility and careless squandering of public funds, they will appeal to arguments to the effect that formation of researchers in basic science -- no matter how remote their subjects are from the realities of the country -- is an essential engine to feed the workforce that will support a robust and technologically advanced economy. Which proposal would you decide to fund?
A final remark, related to the perils of a limited and partial look at science. Studying science and the activities that scientists engage in, is like trying to study the activities of chimpanzees living in a huge dark labyrinth. The sociologists studying chimps turn on a flash light in one lost corner of the labyrinth and observe chimps jumping and screaming, based on that observation they come up with elevated proclamations -- stated with the weight of academic formalism -- about how chimps jump and scream. A second group of sociologists goes to another corner of the labyrinth and all they see are sleeping chimpanzees, so they go and produce the theory of sleeping chimps in contradiction with the theory of jumping chimps. Yet another team ends up in a corner where chimps make noises, thus the theory of noisy chimps emerges. You get the idea, individually none of the partial looks is correct, but if all of them are combined, it may be possible to arrive at a more accurate description of reality. Science and what scientists do is clearly a complex human endeavor. It is not productive to formulate this human activity in terms of uni-dimensional views.