Severous,
interesting analysis - thank you :-) A number of the things that you mention are forum related, which, obviously, I can't resolve with Simutrans-Experimental. The forum-related issues are perhaps a result more of a sort of positive feedback effect: the more development discussions that there are, the more people who want to have development discussions will post, and the less that people who don't want to have development discussions will post, which will further increase the proportion of development discussions. Having a lively development community, however, is a very good thing, since it ensures continual development, which means that the game keeps getting better. Perhaps some thought ought be given to a subforum dedicated to gameplay issues? Certainly, when I have a good preliminary version of Pak128.Britain-Experimental released, I am hoping to start some gameplay discussions about the gameplay issues/strategies particular to Simutrans-Experimental.
A number of the issues arise out of the economics of developing open-source software. Unlike commercial software, things will only be developed if people feel like programming them, which means that programming resources are necessarily limited. Things such as a user-interface for scenario creation, for example, are very resource-intensive and, although desirable in themselves, may not be as desirable as other things on which an equal amount of developer time could be spent (such as network multiplayer mode). A similar point arises with the simuconf.tab settings, although I understand that there has been some discussion about getting a GUI for them. However, the idea of the simuconf.tab settings is that they are set by pakset authors (as with Pak.German), so getting an integrated version of Simutrans with a pakset should, if the pakset is well designed, provide good and well-balanced options. (Incidentally, the simuconf.tab file that comes with Simutrans-Experimental and Pak128.Britain-Experimental (when it will be released) has seperate_halt_capacities=1 set).
Thank you again for your insight - do let me know if you come up with any other things that could be improved (preferably things that do not require a large amount of work, as resources for the development of Simutrans-Experimental are limited) for non-developer end-users.