Hi there - absolutely correct. You are a tool designer and ACAD is still the best all around tool for that type of work.
2d kicks the ass off of 3d for full tool design, however consider the following: 1) You need to do an accurate development and SW (or 3D) is exceptional at doing this. SW can very accurately unfold a part a bend at a time and your config list can show you your bend sequence. You must have this. 2D blank development is just a waste of time and not as reliable. 2) You most likely work with customer supplied models and you can use their models to develop your flats. Very important. With featureworks, you can make dumb models fully parametric with relative ease. 3) The configured part can be patterned and used as the basis for your 2D strip design. You can take all your side views off of the patterned part model. You insert a part into an assembly and then pattern it to your advance. The only issue here is that the part can only be one configuration per station - ideally you want pre & post operation, so you can overlap parts with different configs to achieve this. 4) Parametrics allow you to make a progressive die "shell" with much of the right stuff in the right place. I developed a model that I used to generate a properly timed side view of the stripper, pilot-perf and first two pilots on the correct advance and properly sized - including upper & lower shoe thickness, parallels, die, punch plate, stripper guide pins . . . The timing was also done. From the top view, I was able to manipulate mounting slots and handling holes, change the guidepin style and so on. This allowed me to play with options rapidly without any drafting needed. It saved at least 8 hours a job and it gave me a great "main" side view. I also developed the same for compounds and this could do a basic design shell and project your material costs - great for quoting and so on all with a designed shell as an output. 5) Variational details that you do over and over again but only at different lengths, sizes etc. are really great to do with SW. We used to do a unique type of self releasing form punch (i.e. no ejection, sky hooks, etc) - the same old design but a slightly new length - save yourself an hour each time you generate a fully dimensioned detail. 6) Full blown die design on SW is absolutely clunky. Dealing with fasteners is a pain, all the parametric "fudging" and frankly the drafting is somehow not "clean". Layering is weak, dimensioning a pain and strip layout a nightmare (not too easy to make a concise strip with all those needed "real life" elements - scallop cuts, 45 degree cut-bys, tolerance split for mismatch and so on). There are just too many encumberances to doing a full design with this product in a timely fashion (remember - my opinion only) - it's tough to get a "simplified" side view - my theory has always been to "tell a story" showing just what the toolmaker needs - the "high fidelity" views that SW gives are too cluttered to tell a good story - a good side view can be had, but sometimes you need more views to "tell the story" adequately. Nesting for a stick punch layout for wire EDM? Forget it! 7) Large drawing sets and the need to split your drawings into separate documents is a barrier to sharing data between your drawing panels. Not impossible, but another encumbrance. I'm personally used to a single sheet "monolithic" drawing with many frames scaled up or down as needed. Exchange of data between sheets is easier with raw 2D - easier to "cheat" which is sometimes needed. 8) On the upside if you want to do full 3D designs, there is no CNC prep down he road and your data integrity is absolutely awesome. This is the upside and you can easily see the relationships between components. In come cases, this is better, sometimes worse. With 2D I like to do a superimposed design with layering viewed thru the upper. I can see all of the design at a glance and use layers to see different states of the design. 3D can do this as well, but sometimes not as easy to see things. I have used 3D at times to develop a complex forming operation - good for visualizing a design. In one case, I had a 3 sided form op that clasped to the upper (classic - formed around the upper, never to be removed), so I needed to design a form punch with the ends that moved out on the downstroke and released on the return stroke - the good news - the part sat on the lower pad and did not clasp the punch - 3D helped a lot there - but the base design was still 2D. 9) Libraries can help in either realm. Developing a good 3D library takes time and will make 3D design easier. Most likely, you have a good 2D library that is already saving you time. Personally I find this to be the thing that I miss the most in 3D design (tooling specific). Maybe that's one of the reasons that it is "clunky" for me.DISCLAIMER: The preceding is simply one person's opinion and this being offered does not preclude others from doing it better or having great success with pure 3D tool design.
For me, your original statement about the economy of 2D design (for this type of work) rings very true.
Later,
SMA