this topic is of interest to me. Obviously the portability is key, but unless the subjects are static, they need some degree of speed also. It argues for some form of auto exposure & possibly autofocus. The little retina qualifies well for portability, but falls down on the speed part. I'm slow on rangefinder coincident focusing, but even if fast, its not going to catch the action/people shots.
It really comes down to a workflow and subject matter issue. What are you shooting, how are the conditions changing, what is going on, etc?
One of my friends does street photography with a medium format TLR with waist level finder, and most people barely notice him using it when he takes shots. The thing isn't what you could call small or really concealable, but by adjusting the straps properly to hold it against his body where he can casually look down into the waist level finder he can hold his phone in one hand near it while adjusting the camera and tripping the shutter with the other hand. He is just another person staring into his phone and not paying attention, so no one pays attention to him...
If he keeps working in one area with the same lighting, and keeps photographing people at the same distance... Well there is no metering or focusing to be done, and he can fire away from the hip using a very manual camera.
Humans often forget that the key to hiding something isn't for it to be unseen, but rather for it to be
unnoticed. A point that is a rather small but extremely important distinction.