No arguments to any of the above, aside from the fact I'm looking at the comparison from the standpoint of a CAD adiministrator, not a dedicated programmer. Most engineering companies don't or won't have a dedicated programmer on staff and put the AutoCAD customization onto the same person who is responsible for...
- knowing all the little corners of AutoCAD, including known bugs and "gotchas"
- setting up and enforcing standards
- data entry for all the programs (for my part, that means piping specs, data files for piping, steel, electrical programs)
- adjusting the support system to handle differing requirements from various clients
- basic support for supporting programs such as Word and Excel
- evaluating new software and new releases
- planning and executing upgrades and licensing
- knowing the basics of the profession (I need to be part stress engineer, part materials engineer, part piping designer, part civil/structural designer, part IT professional, the list goes on)
... and a bunch of other miscellaneous stuff. Not a lot of time left over to be learning a dedicated programming language such as C#. The only reason I have gotten as far as I have is with previous programming experience with C, Java, and machine code (after which LISP is a breeze). Trying to justify to management the hours spent on converting all of the existing support code would be a hard row to hoe as well especially with a downturn in the economy and the reluctance to spend money for little realized returns.
The problem as I see it is one of time management and diminishing returns. As both AutoCAD and programming tools increase in complexity the amount of time required to be proficient at both increases as well. Eventually a "critical mass" is reached where people can only be proficient at one or the other. I'm sure we would have no problems finding .NET programmers, but their knowledge of the ins and outs of AutoCAD *and* the industry we are in *and* .NET programming as it applies to AutoCAD? Highly unlikely - it took us 3-4 months just to find somebody with sufficient AutoCAD experience (not even programming) to back me up. In fact, I see this as the reason why AutoCAD products are on a slow decline in terms of useability. The programmers that write the software do not have the time to gain the industry experience of actually using the product, while those that are very familiar with using the product in the chosen industry don't have the time to get proficient at the programming tools used to customize AutoCAD to the nth degree.
I still maintain, again from the standpoint of a CAD administrator rather than a programmer, that for most of the common automation tasks needed by engineering companies LISP is the better tool in terms of time spent developing, deployment, and upkeep. .NET is better for large, full-service-program type automation where the company is unwilling to spend money on a product developed by a software company or no product is available.