– Horizontal takeoff lowers gravity losses but increases drag losses. Aero plus gravity losses for airbreathing launch vehicles are generally larger than VTO rockett vehicles.
Yes but note that drag is typically 1/10 that of a gravity losses. In the case of the Saturn V it was 1/38 that of gravity losses. IMHO VTO LV designers should be asking “How did they do that? It’s tiny“
So doubling drag losses and halving gravity losses would be a very good trade. That delta V ultimately translates to lost mass to orbit, a physically smaller vehicle or a smaller propellant load. Whatever way you want to play it.
I’m also extremely wary of unqualified statements in this area, especially around SSTO systems but of systems in general. 
People make statements that only apply apply to VTO rocket systems. These are then read as applying to all systems, even when they don’t.
Airbreathers spend a lot of their time at high q overcoming drag.
Regretably that would be one of those unqualified statements I’m talking about.
I’m guessing (pending any qualification by yourself) you’re talking about SCramjets :(which do so to eke out the very poor T/W ratio. I’ll qualify that by noting that I don’t know if people quote T/W ratio on a SCramjet I don’t know if they include the mass and fuel consumption of the accelerator engine that get it up to that. I don’t think they do though. Last time I checked they could do 4:1, which doesn’t sound bad, but again what’s the mass of the accelerator and its fuel consumption? IIRC the J58 for the SR71 was about 5.8:1 (IMHO not bad for early 50’s engine tech) but it could do so from a standing start.
Skylon and SABRE were designed to avoid depressed trajectories, which give the prolonged time in the atmosphere you’re talking about. Which is why when the cost of airbreathing (in terms of drag through the nacelles) exceeds the delta v gained it switches to rocket mode.
Isp (thrust/propellant flow) is not as important as effective Isp ((thrust-drag)/propellant flow). Total required delta V is closer to 10000 ft/s if I remember right.
John
True. And AIUI the ability to predict the drag at these speeds appears to be significantly poorer than at lower velocities, as is the ability to predict SCramjet thrust at those velocities. Rockets OTOH have quite predictable thrust levels whatever the velocity of the vehicle carrying them.
BTW I ran the delta V losses for all those launch systems and in all cases it comes to <6000fps. Not theorectical vehicles, but actual vehicles built and flown (although all VTO rockets). Historically I’ve used 9200 m/s for LH2 fuelled vehicles and 9100m/s for all others. The surprise was the Atlas 1 needed that much. The Delta 7925 was really outstanding in this regard, but by firing it’s 9 SRB’s in 2 groups (the 2nd with high altitude nozzles) it effectively became a 3 stage vehicle.