As a natural scientist you are disposed to believe that there is something on the outside of your consciousness, and also, that you as a human being have suitable means to investigate those things.
And there is the theory of critical rationalism and the praxis of how geoscience (and palaeobiology) is really done.
So what about the chronic underdeterminedness of geoscientific (and paleobiological?) models: Is the remaining uncertainty large enough to consider much of the geoscientific knowledge as mere constructs which are (forseeably) bad descriptions of real things?
The following list of sources of uncertainty and of other problematic points is without order, representing a collection of thoughts I had during one of those late night train journeys:
(1) Uncertainty from the misunderstanding of recently active complex processes (e.g. sediment transport by a river) used as model systems for ancient processes.
(2) Uncertainty from a lack of knowledge about the longterm consequences of currently active processes.
(3) Anactualistic processes (e.g. on Hadean earth; recovery after a meteorite impact).
(4) Selectivity of geological records and the variety of factors by which it can be governed.
(5) Equivocality of deeper earth investigation methods (geophysics, "1D" outcrops from drillings).
(6) Problems arguably associated with the "descriptive tradition of geoscience", which passes along varying ideas about the necessary exactness. There can be a certain disregard for matters of decision making, the question of how to choose a favourite hypothesis, the distinction between noise and information relevant for a problem.
(7) Plurality of fieldwork procedures [and other aspects of methodology], i.e. documentation methods in the field and the decision makings/ usuals paths of inference involved. Enhanced by national/ language boundaries - there can be different schools, sometimes failing to communicate.
(8) Choice of model systems: Due to the complexity of the matter we are choosing well understood model systems, but the comparability is overestimated in the specific case.
(9) Overuse of a canon of "inherited" procedures, which lead to systematic misunderstandings or put constraints on thinking (e.g. the drawing of sedimentary logs).
(10) Metamodels and metatheories (e.g. global palaeoclimate models, supertrees) may be particular problematic in geology/palaeontology as historical sciences. There may be so various data and multiple steps of inferences that circularities are hard to avoid and/or the question is, of how much value the insight from the model is.
(11) Bad integration of data in multidisciplinary approaches, if the model assumptions on which the integration is based are not well considered.
(12) Overinterpretation as a notorious phenomenon in geoscience, coming from the chronic paucity of data, the community's failure of encouraging modest/ honest claims.
(13) Unsystematic or right-out defective handling of temporarily or principially not-available data. (Hypothesis hinges to a considerable degree on the missing data - though you as the inventor are saying the solution of the problem is only a matter of effort.)
(14) As in other sciences: the dealing with falsifying data. As your object is so complex, you as geoscientist may always find an exception demonstrating that you are not (so) wrong.
(15) Underuse of quantitative approaches, another relic from earlier times of geosciencemaking.
(16) Changing accessibility of important outcrop areas (but also archives) limits falsifiability of hypotheses.
(17) Misinterpretations from reducing/ enlarging the dimensionality of a problem in an inference step (e.g. reconstructing a time series of 3D models from a number of 1D sedimentary logs). Not always considered: There is not necessarily a proportionality between the time span of a process and the volume of "waste" it creates or deletes.
(18) Downscaling, if done without reflection/ definition of the way of doing, can reduce the reproducability/ comparability.
(19) Thinking in cyclicities and the tendency to infer cyclic processes although their support is poor.
(20) Misconception about what the data are and where the interpretation begins. One might think this is only a beginner's mistake in geoscience - however, it may happen as well on a higher level if you have multiple stages of inference.
(21) Thinking in categories as though they represent something natural which is not defined by the observer (e.g. genus, family in biological systematics; stage, era in chronstratigraphy).
So, look critically at your own models in geoscience (and paleobiology). Are they well constructed?
Shcherbakov, D.E. 2008. Insect recovery after the Permian/Triassic crisis. Alavesia 2: 125-131. [pdf]
Shcherbakov, D.E. 2008. On Permian and Triassic insect faunas in relation to biogeography and the Permian–Triassic crisis. Paleontological Journal 42 (1): 15-31. [pdf]
The Alavesia paper outlines a three phase development of Triassic entomofaunas, beginning with (I) a low-diversity episod of P/T recovery dominated by Paleozoic insect groups, followed by (II) a summit phase with typical Triassic taxa in the Anisian-Carnian, and, with a decline in diversity, ending in (III) a phase dominated by Late Mesozoic elements, especially featuring new aquatic insect groups. Shcherbakov suggests, that each of the transitions began in the humid low latitudes and occurred later in the higher latitudes, i.e. the boundaries between those three stages are diachronous.
In the Paleontological Journal paper Dmitry Shcherbakov looks at the insect diversity of Late Carboniferous to Triassic localities, counting the proven occurrences of insect families per stage ('stage' as a chronostratigraphic unit). He illustrates the change in aquatic/ terrestrial, phytophages/ predators, modern/ ancient groups and explains the ecological, evolutionary, and biogeographic background of diversity fluctuations.
Shcherbakov, D.E. 2008. Madygen, Triassic Lagerstätte number one, before and after Sharov. Alavesia 2: 113-124.[pdf]
This review paper begins with a recount of the research history of the Madygen Formation as a Triassic fossil locality, beginning with the geological fieldwork in the 1930s (by Kochnev) which led to the first finds of a fossil flora, classified as Triassic, and to the introduction of the Madygen strata as a separate stratigraphic unit.
In detail the role of paleoentomologist Alexander G. Sharov is recognized, who lead five field expeditions between 1962 and 1966 to a fossiliferous point in the northern Madygen outcrop area (Dzhailoucho). These campaigns turned out as the most successful with regard to the number of recovered insect specimens and other fossils. The historical part is followed by a short overview of the flora and non-insect fauna.
The main part is a synopsis of the particular insect fauna of Madygen. Besides the exquisite state of preservation, several figures illustrate why Madygen really is a lagerstätte: Members of twenty insect orders and 96 out of 106 insect families known from the Ladinian/Carnian have been reported from the Madygen Formation.
In this order beetles, cockcroaches, and homopterans represent the most abundant groups. Among rarer groups are certain specialities, such as the most diverse assemblage of titanopterans. Modern insect orders are represented by several groups of early dipterans and the earliest hymenopterans (belonging to the group of sawflies).