I think the problem is more possibly where you are placing emphasis rather than the points themselves.
Yes, a +1 is a big deal at the outliers. But its not a big deal as you drift away from those edges.
You pretty much nailed my read of their argument... Pat was putting so much emphasis on the outlier effects while ignoring that almost NO modern game (i.e. the types where you're most likely to see "builds" and planned optimization) build their system to have lots of outliers because, frankly, extremely low/high probabilities of success aren't actually that interesting.
Rolling 10-20 times just to hit once (because 10% doesn't mean you'll hit once every ten swings, it means you'll hit, on average, once every ten swings, but with the flat probability of a d20 there's quite a bit of swing until you're looking at hundreds of swings) is no one's idea of a good time.
Outside of TSR-era D&D (ex. 1st level Fighter with a Str 17 vs. AC 2) you just don't see odds like that too many places.
The fact is that most modern systems seem to be much higher than that... 50%, 60%, even 75% odds are not uncommon in modern systems (my system, for example uses a baseline for combat of about 60% or about three hits during a five round combat... this was also the baseline for 4E).
5e makes hitting even easier... low level opponents with ACs of 11-14 are pretty common while PCs probably have +5 to hit out of the gate for 60-75% odds of success with each attack. You'd have to be playing an 8 Str wizard using a non-proficient weapon to even get down to a 30% hit rate vs. AC 14.
And in those ranges, +1 is pretty trivial.
That's why I considered Pat's arguments off-the-mark/not communicated well. Sure, everything they said about the probabilities they mentioned were accurate, but almost no system where optimization is a big deal actually uses probabilities in the range their comments were primarily focused on.
Yeah, +1 in their edge case was a 50% improvement from 10% to 15% odds of success. But if situations where your odds are as low as 10% come up only once in a hundred checks... what's the point of using that 10 becomes 15% is a 50% bonuses in your argument. It's like arguing the utility of tiger repellent when travelling in India to someone who spends all their time in rural Montana. Sure, tiger repellent is useful in India... but its also a non sequitur to probably anything the Montanan may have been talking about.
Heck, from playtesting my own system +2 situational modifiers were largely ignored as not worth tracking by most players and skill-based boons (basically non-combat feats) needed to be about +5 to check results for 2-3 interrelated actions) before they were seen as worth taking on par with other options (adding graded successes that scaled up for every 5 points the base TN was beaten by also helped too since +5 gave the PC with the boon the same odds of pulling off the next higher grade as a person without the boon had of getting a basic success).
EDIT: Another thing I really think needs to be distinguished is that there were a lot of CharOps builds that were never intended for play; but rather theory-crafting exercises to see what they could break. A lot of those had critical weaknesses that were ignored because they were attempting to build for one specific target (best damage, best defenses, highest hit points) even if it gimped them elsewhere.
Those same people almost NEVER built the PCs they used in actual games that way because they understood that there are things like diminishing returns and that its not worth utterly gimping something else important just to chase them (i.e. I've never seen the people I knew who frequented various 3e/4E CharOps boards to consider taking a -2 penalty to all your saves/defenses as being worth getting just a +1 to hit).
Basically, you need to distinguish between theorycraft optimization and actual in-play optimization (which was rarely as severe unless it was someone copying a theorycrafting build from the boards without realizing it was a theorycrafting build... which usually left them with some gaping hole somewhere).