I was referring to the guy from TastyTool adding flies inside the condor. It's not a hedge. He's adding risk. OK, so he can reduce the local delta pos, but he loses double outside the wings. I don't want to quibble about the actual PNL.
Long wing-spreads act as synthetic long puts and calls when traded with the underlying. Say you're trading XYZ and long the DOTM strangle. Long XYZ and the put = synthetic long call. Short XYZ and the call = synthetic long put. Or you're trading defined-risk as a futures trader with the long strangle. ATM condors (flies) are typically 1:1 risk so you're swapping risk for terminal performance, i.e., The prob of expiring outside the wings of the strangle may be 10%, but the risk may be 10:1 against.
And yeah, I am referring to the strangle and not the IC, but the principle is the same. The condor is simply there to define the risk.
I sell wing-spreads all the time, but against the underlying.
Damn. I actually got almost all of that. I must be getting smarter from associating with you folks.

So... OK, he's reshaping his original risk profile to reduce the local delta position, and paying for it in increased risk elsewhere (and the overall risk increases because, well, ain't no free lunch.) I can only assume that you're saying it's mega-stupid because the increase in risk he's taking on is disproportionate - which presumes that there's a smarter way to do this. I may be mistaken, but it looks like you're saying hedging is it.
I'm still quite clumsy with synthetics, although I have the basic idea down; no "natural feel" yet for how to use them. That's likely a large part of my problem in understanding/visualizing adjustments. But I can see applications to hedging from what you're saying there: if I have a call that's being tested, I can convert it to a synthetic put by shorting stock. ...OK, I'll have to burn some midnight oil; I need to see the effect of this to really nail it down. What will price movement do to this trade if I short enough to zero the delta? What about 10% more, or less? Grrr. Wish I could visualize this stuff already.
And sure, I'm familiar with the risk/reward tradeoff; 10% ProbITM implies an inverse risk ratio (although, just off the top of my head, I seem to recall a roll-off as you go further OTM; i.e., you pay for the "safety" of being FOTM.) The best RRR is still ATM, or somewhere close to it - but this seems like it would require the ability to make directional predictions. Lots of questions in my mind about wings - tail risk vs. return vs. stops vs. hedging - but that's a different topic.
(Perhaps this is a bit of craziness on my part, but in my Copious Spare Time, I'm slowly starting to build an ML system where I can run different models and test my assumptions. One of my former employers gives me access to a distributed system environment with DS/ML tools, and I'm pretty good with the relevant Spark/Scala/Python tooling on it - used to teach the stuff. 'Cause, you know, I don't have enough to do, and too many free hours in the day.)
By the way: I've read quite a number of your threads/journals here on ET, and have been absorbing whatever my current (expanding, but too damn slowly) capacity permits. It's been one of the most effective increments to my education about options - I've stared holes in my screen looking at your trades and trying to understand everything I can about them - so I'm quite grateful. If we ever run across each other somewhere, the beers are on me as long as your kidneys hold out.