The author, one Michael Duffy, begins by relating his personal experience with Greenpeace. He was a member of the organisation until they asked him for money with the claim that 30 000 species were going extinct every year. When he asked them where they got this figure from, they couldn't come up with a good answer and he quit. He goes on to criticise the method of extrapolation used to come up with such estimates of extinction rates, and then extends his criticism to include models used to predict climate change. He gives various examples of models that didn't work, and quotes various experts saying that inconsistencies appeared between models and how things really played out. In short, he's saying that the method of computational modeling tends to over-reach, is being used to come up with policy without sound proof, and is largely a house of cards.
He concludes his article by condemning a whole branch of science:
"We often hear that the predictions accepted by the Intergovernmental Panel on Climate Change are based on "the science". It's important to realise that this is often a very different type of science to other science, the sort that explains why a jumbo jet won't fall out of the sky or why a certain treatment will cure a certain disease."
His false dichotomy of the sciences: woolly-headed models versus hard and concrete laws, falls flat in his choice of examples, which betray his lack of knowledge of how things work. It is precisely computational modeling that is used to design jumbo jets. Engineers are unable to come up with a simple explanation for how jets fly. Planes are designed not with a few computations on paper, or even by building simulations on computers, but by building models and putting them in wind tunnels because that's the only reliable way to figure out how a certain shape of airframe will behave in the air, precisely because our knowledge of fluid mechanics is insufficient to allow us to predict this reliably from first principles. By Duffy's reasoning, then, we shouldn't fly in planes because they are designed with the use of models, which is not 'real science' in his eyes. One can point to examples of planes which failed or almost failed, like the Comet airliner of the '50s and '60s, to support this foolishness.
Bringing the issue back to the environment, we cannot deny that there are species being lost. The numbers thrown around by NGOs and environmentalists vary, and personally I feel that they should be more responsible with where they get their numbers from and their sources' reliability, but species are going extinct and there is empirical evidence for this. Brook, Sodhi, and Ng's paper in Nature (424: 420-426 (24 July 2003). doi:10.1038/nature01795) for instance used species checklists over the period of Singapore's colonial occupation to the present day to determine what used to be found here and is no longer present, and from these known records extrapolated them to estimate species losses for groups of organisms which were not so well documented. But even if we ignore the inferred species extinctions ('inference' being the concept which Duffy finds so difficult to understand or accept) the recorded extinctions alone are high: more than 40% for mammals, 25% for plants, 35% for butterflies.
Furthermore, it is a logical fallacy that Duffy commits (commonly known as 'association by hand-waving') when he segues from taking issue with extinction rates to questioning climate change; as if they are both part of some great conspiracy by environmental groups to misinform the public. This gives environmentalists credit for more public and policy influence than they do have. Duffy says that we should not trust these predictions,
"...because virtual science is ripe for manipulation, usually unconsciously, by virtuous scientists. Few people are aware of the large element of subjectivity, not only in the design of immensely complicated general circulation models, but in the data that goes into them. Even basic information such as contemporary temperatures is often incomplete or uncertain and tweaked by those who operate the models."
There is not much that one can say in response to this, because it's not about evidence any more but whether one trusts the evidence and those who gather it. Well, if one cannot trust the predictions, one should at least trust the observations. If one yet refuses to trust observation, then nothing will satisfy one's critical eye.