To me, the frustrating thing with modeling is the IR/mic'd cab piece of the puzzle. I still don't understand why we can't just have a modeled sound without the mic, similar to what Powercab accomplishes. Quite honestly, Helix with Powercab is the closest thing to an amp in the room set up I've played, next to real amps of course. Why can't the sound coming from the guitar speaker be modeled without the mic variable?
I gave modeling such an unfair shake in the past, and then I tried my amps through a Suhr Reactive Load IR. I was SHOCKED at how close to modeling my own amps sounded. The IR variable sounds much better than I'd anticipated, but it still gave my amps that modeled sound and feel that I always sort of hated. Once I heard my Two Rock Studio Signature through an IR, it clicked for me. This is just what it sounds like to mic an amp and listen through monitors, headphones, or PA speakers...modeled or otherwise. Once I accepted it, now I can just sit back and enjoy and not obsess.
Because that sound depends so heavily on a lot of factors. The room, the volume, the placement of the cab, the speaker cab used, where the listener is in relation to the cab etc. People have tried replicating the sound of a cab with all kinds of filters without success. Impulse responses solve much of the problem with the caveat that it is the sound as perceived by a microphone. Even if you use a reference microphone that has no coloring of its own, it will still not be the same thing as our ears perceive. Most of us are used to listening to a cab blasting at our knees. Maybe eventually someone will figure out how to properly model a guitar cab or DSP improves to the point that it is possible.
I would recommend everyone tries a tube amp through a reactive load and IRs before saying modeling does not sound like a real tube amp. The results are often indistinguishable. That's the sound you hear through a PA at a big gig, that's the sound you hear on records. The sound of a guitar amp played through a cab that is then miced and mixed.