Is choosing a new modelling project like selecting a car or computer where you can select a range of options or optional extras? I want one with this, with that, I would like it to do this and be able to to do that. In which case this combination is the best choice. Or are we more fixed, by our selection of prototype, scale, track gauge, availability of stock etc? Availability of space and cost must also be a factor, even if the cost is not always the first consideration.
I have been thinking about these things because my attic is filled with a decent sized 00 layout that allows me to run my BR blue era diesels. The trackwork was Peco because at the time I either had to get something built quickly or I wouldn't get anything at all. If I say the layouts gestation period was 9 months you can probably work out the significant event that would impact on available modelling time.
In an ideal world I would go into the attic one day and everything would have been (tidied up would be good) reset to the default setting. That is a blank canvas. If it was, my choices would be very different now. But if I had that blank canvas would I work in P4? I have been doing some modelling in P4 for over 20 years and I am currently getting a little scenic diorama ready for a local show. But would that be my choice for a large attic layout. The honest answer is that if the larger steam locos were easier to convert then I probably would. But I'm actually not just talking about ease either, the cost is a factor. With RTR locos to around £100, the thought of spending the same again to re-chassis and re-wheel is off putting.
The thought here then is perhaps these choices should not be fixed by pre-conceived fixed factors. I am a P4 modeller, therefore I cannot work in anything other than P4 is perhaps not the right answer - at least for me, but I do accept that this is a personal journey and that may not be the answer for others. My logic is taking me more to the view that "I have 36' x 9' to work with and I want to model the X railway in 19 oatcake, therefore the most realistic prospect of me achieving that is to do it in this way."
A large sized layout would just not be achievable given the time I have available unless compromise was made of track standards and the use of RTR stock. Does that make it a less valid exercise in representing a prototype? I am more convinced that it does not.