If your goal is way off the race estimator then you need to tweak your training. If everything looks about right there is no need to make big changes.
I, and nearly all hardened experienced marathon runners, definitely disagree with you greatly on this one.
I would actually strongly say that for most 'AG runners' (nonelite), 5ks-HMs are almost interchangeable and accomplishable just by modifying pace and race strategy, while using similar training volume and intensity.
The MARATHON, however, is a totally different beast. At 26.2 miles, you better have lots of high mileage training in your legs to keep the distance - the VAST majority of folks underperform their 'calculator' estimate by 20+ minutes the first time out, and 10+ minutes on subsequent attempts.
Even if you can run a 70.3 at a good clip, and feel fine the next day, you will be in for a monster rude awakening when you show up to marathon day, expect to run an adjusted 'calculator' pace based upon even a recent standalone HM performance, and then watch in horror as your legs cramp full out at mile 20.
For the OP, I def think you can FINISH the marathon on 70.3 training, and the 'respectable' component is totally subjective and up to you. If you're content with finishing +30 mins slower than your calculator prediction time, you'd likely be fine. For <5 or even <10 mins to your marathon calculator time, highly unlikely, and you may very well get a short-term injury during the race that will sideline you for a month as a result.
I will note that here on ST, there are a surprising number of fast people that can run sub 3:15s on <25mpw. (There is some guy who runs sub-3 on <15mpw!) These guys are the EXCEPTION to the rule. I've been on plenty of marathon forums in the past, and nearly nobody pulls this off (although mannnnny people try.) And it's not just my opinion - there will be virtually no establish triathlon coaches who will claim to be able to reliably get you to a strong marathon performance of pure triathlon training - they would actually dissaude you from the marathon until you can commit to it.
So while I totally agree with you that the marathon is a different beast entirely from shorter runs, I am not sure how much of an exception to the rule is that you cannot do reasonably well on "less" run training than a typical marathon plan would have you do.
I don't personally feel I am in any way a special athlete. Just a guy in his late 30s who trains a metric ass load. I have recently run a sub-3 marathon at the end of a season of Ironman training, after spending about 6 weeks in more of a run bias before the marathon. Even now as I am finishing up the build for Boston coming up, I have been generally averaging 15ish hours a week, of which the running averages 35-ish miles (longest run weeks in the low 40s).
The OP suggested basically taking the peak fitness of a 70.3, adding on a number of weeks of a slightly more run bias before the marathon. Is that going to yield an absolute best marathon time? No. Of course not. Would it be pretty close? Within 5 or 10 minutes of a best possible time? Probably.
There is definitely a running durability issue with running long that needs to be trained. But much of the fitness component can certainly come from triathlon training also.
I don't disagree with you, but the main points:
1. You are a hardened IM athlete training 15 hrs a week for a long time. OP is a 70.3 athlete, with presumably no or remote IM experience. BIG difference between IM volume commitment and 70.3. Sure, you CAN train as hard and harder for a 70.3 than an IM if you so choose to, but I'll wager the vast majority of AGers who aren't national class train a lot more for the IM.
2. It's hard to add enough run conditioning in <2 months to go from a max long run of 13 miles to 20 miles without a high risk of injury or burning out your legs for race day, even if you are 70.3 race-trained when starting the transition. At least for IM, you're doing a few runs over 14 miles, if not more than 18 miles; in 70.3, you never do that run, and you might even max your run distance out at 10-11 miles as in a lot of plans. That's a lonnnnng way to go for 26.2.
3. I will wager you are a lot more talented than you think. To compare, I will compare myself to you, where I have done several blocks of nothing but pure run training of 70+mpw with quality (Pfitzinger's Advanced Marathoning 70mpw and up plans), and never gone sub 3:10, even in my early 30s. There is no way I can run faster than my peak back then, and I still could not run as fast as you. And I'm no super-lame ability person either; I walked into those Pfitz plans running a mediocre but not-slow 20:xx 5k, which is already faster than most AG triathletes can run standalone 5ks. (Got that down to low18s with all that mileage.) I also ran a flat HM after a strong training block 70.3 for myself where I couldn't go faster than 1:34, and I know for sure that time would have slowed catastrophically had I attempted a 26.2. So I suspect you actually are one of the people that runs a lot better than the average AG triathlete on less miles trained. (Still give you props for your 15hrs/wk!) I'm going to guess you're one of the fastest in your AG even at big races and are USAT all-american?