So what about model railroads? Here's the equivalent for my layour:
What went right?
* Double deck layout This layout's turned out to be the perfect size for me -- large enough to feel like I'm going somewhere, and small enough for me to actually build and maintain it. After a series of shelf layouts, I'm very, very happy.
* Staging to staging + reverse loops The loops for turning trains have worked well in operation. I only wish I had a continuous run option on the lower deck when troubleshooting as I'm not really willing to run trains unattended on the (unscenicked) upper level.
* No yard Most model railroads have a freight yard both for the fun of switching, as a centerpiece of busy action, and as a place to show off all their locomotives. By consciously giving up the yard, I gained a lot of space -- probably an extra town or two. As a switching layout, the extra industries make the layout much more fun and realistic. Finally, any attempt at a yard would have ruined the prototypical aspects of my layout as there was no way I'd get anything resembling the San Jose yards in my garage.
* Model based on maps and photos Jack Burgess is right -- you can get wonderful, realistic detail on a layout when you match a prototype and work from photos. This is a lot like preparing a movie set. I don't know how Hollywood set dressers fill a teenager's bedroom on a movie set with appropriate clutter, but I can't do it on my layout. The San Carlos St. overpass is a nice example; I'd originally designed it as fill with a short bridge. However, one day I realized the prototype was a viaduct. Once I switched over to the viaduct, I also realized I could (prototypically) keep the old pre-viaduct street that survived next to the bridge, and have enough space to add the packing house's boiler and oil tank. I've also got space for a (probably prototypical) hobo village under the bridge.
* Just build it I'd been hesitating on building this layout until I could prepare the garage appropriately, put in lights, etc. I finally decided I just had to start building. I'm glad I did, for I'm enjoying the last couple years of construction, and I'm getting skills and ideas for the next layout.
* Focus on my skills On previous layouts, I learned (the hard way) that I was good at electronics and scratchbuilding structures, but I'm no good at handlaying track. The layout got flex track and Shinohara turnouts, and I got the layout running. I also knew I could do some unique buildings, and I could do them over time. This meant that I could do a big industrial scene without going nuts or spending a fortune on structure kits.
* Treat it like an assembly line Some of my modeling skills still resemble my teenage years -- using whatever wire color was available for a circuit, buying electrical switches piecemeal to keep costs down, and trying to make do with the inexpensive option. I realized early-on that wouldn't work if I wanted to make progress on this layout. Approaching each problem -- laying track, putting in power, or attaching switch machines -- from scratch meant I didn't get to get practised at any one task nor did I save time by building several assemblies at once.
I found it much better to place an order at an electronics supply house for enough toggle switches for the entire layout rather than mix-and-match whatever was available at Radio Shack on each shopping day. I bought Tortoises in bulk so I could prepare several at a time and install switch machines when I was in the mood. I preprared the Tortoises in assembly-line fashion, preparing several for installation at once and soldering wires on them on the benchwork. When I installed them, I'd waste any excess wire, but I significantly sped my move to powered turnouts. With a layout larger than 4x8, these kinds of work practices really speed construction.
What Went Wrong?
* It might have been nice if the layout were level. Okay, maybe over-engineering does have benefits. My layout has few legs and big unsupported spans so we can use the space under the layout. I also didn't fully plan on how to support the upper deck. As a result, the benchwork is a bit overloaded and I've got some sagging in the layout. Between this, the sloping garage floor, and a lack of levels during construction (apart from a single 24" level), the layout and grades didn't get as level as I would have liked. I wish I'd gotten a laser level before I started; the 30 dollar one I got at Sears a couple months ago would have been a *BIG* help when building the layout.
* Not enough staging. Staging tracks, fiddle yards, and yards for switching are all used differently, and require different sizes for the same railroad. When I started treating my staging as a real yard, I found I didn't have anywhere near enough space.
I'd based my staging on the number of trains I expected to run during an operating session. However, I quickly noticed that because the staging's exposed, I could treat it as a yard and sort boxcars between trains. This worked fine at first, but as I got enough freight cars to simulate the traffic, I found that four tracks was way too few tracks to do a proper job sorting. more difficult.
This was made doubly worse when I switched from car cards to a computerized switchlist that tries to model the movement of foreign cars. Not every foreign car would get used in every operating session, so I'd end up with a fair number of cars that don't leave the yard... and get in the way. If a yard is staging, I might only need one track per train (plus a spare), but if I'm switching or storing, I need space for the cars that aren't moving, and extra space to speed sorting.
* Garage layout. My layout isn't well sealed from the elements. As a result, the track gets way too dirty between use, and the reliability of the layout isn't where I'd like.
* Ground throws for throwing switches. When I started the layout, I assumed that a switching layout meant I could live without powered turnouts. I thought I could instead use Caboose ground throws. Between the garage environment, small locomotives, and my higher expectations about making the layout reliable enough for operation, I quickly figured out they just plain wouldn't work. I had lots of problems with electrical contact and with switchstands coming loose. This year, the railroad did a massive investment in Tortoise switch machines to improve operation.
So what's on your "What went right? What went wrong?" list?