Preamble

The Alternity Cosmos II is a complement to a dice role-playing game that uses heuristics based on hard-science to 'build' plausible star systems for the Alternity game: http://www.alternityrpg.net/resources/1375/original/cosmos-2.pdf

The algorithm* itself is the center of the current question, as I would like to know which are it's shortcomings, other than the quantization of the results product of representing probabilities with discrete dices. *: Understood as the steps and assumptions that those steps make to end up producing a plausible star system from randomness.

To answer the question this 'game specific' detail is not useful, in fact counterproductive, when analyzing the correctness of the ideas. The following website can ease the main pain dices can cause when reading the document: http://anydice.com

The other main shortcoming of the algorithm is that it uses the GRAPH method as magnitude in many cases. There is the equivalence to real world metrics: http://www.warrensburgweb.net/alternity/system/GRAPH.html

Question

What does the logic of the Cosmos 2 'algorithm' lack in terms of plausible science and considerations taken into account?

Which would be a right way to predict it with the information known up to that point in the algorithm?

I'd like to know what does it get blatantly wrong, what does it ignore and what does it get partially wrong. But only in the cases where it pretty much destroys the whole model. Things like: the model does not take into account solar winds, and at this distance, planet N will be stripped of its atmosphere. (Pretty big deal! The precise example is already covered well enough, I think.)

An answer pointing something out should also tell how the conclusion is met, either by an existing formula, a real life example OUTSIDE EARTH (biosphere chemically changes everything) or with a simplification good enough using the parameters known. The formula can either be actual science or a plausible assumption that fits our current understanding of the universe.

Off-limits

There are some details that are not covered in the algorithm, and thus are arguably shortcomings of the model, that I'm not at all interested in including.

I'm not interested in:

• Strange orbits other than the ones depicted in the original text.
• Oort cloud details or Kuiper belt like asteroid belts.
• Trojan belts or Trojan planetoids (other than the already detailed on the document).
• Topological details of the surface of any orbital body.
• Biosphere details, assumptions or considerations of allegedly habitable exo-planets.
• Basically anything that does not concern the formation of an orbital body or it's stable distribution around a star.

Other details, which I already know are not covered well enough in it, but I'd like to get answers about:

• Better atmosphere content prediction that includes chemical composition, not the shallow GRAPH system. (when it exists)
• Better hydrosphere content prediction. (when it exists)
• Better surface composition. (when not talking about gas or ice giants)
• Core composition on planetoid and lesser bodies (< 900km).

What is the use of such knowledge?

I'll implement programatically the algorithm, changing the 'dice throws' by real probability functions, and including all the corrections that we end up adding as answers to this question. Woho! Free plausible star systems for everyone! A net win to all the world's daydreamers.

Answers

Feel free to post a partial answer, i.e. only addressing a single wrongness that you know of. I will accept the answer that gives either of two: The most corrections to the given model or the most complete answer to a deep and impactful wrong assumption done by Mark Peoples, the author of Cosmos 2.

The hard-science tag is not gratuitous, and I will not accept implausible made up things or dodgy assumptions, independently of the awesomeness, amusement and hilariousness they provide.

Note: I know this is BROAD! But the answers are very specific.