In MY part of the country, those massive gain antennas are fairly useless because they simply hill top
So they cause plenty of interference to other systems a hundred miles away yet have poor LOCAL coverage due to their gain
In the hills and valleys of Ohio and West Virginia I've found that 6Db gain is the max for this terrain and many commercial dealers are using 3db repeater antennas to fill the valleys better
Without down tilt, most of that antenna's gain is being wasted at the horizon
I have replaced MANY 9db antennas with 6db antennas and LOCAL coverage was better every time
I absolutely agree with both of your points, Ralph. My home is at 6,250 feet 2 hours east of Los Angeles. Our primary repeater is at 6,450 feet just up the hill from me. We are top-of-tower mounted. The back-up repeater at my home is side-mounted on an 85 foot tower and kicks butt. Better than the primary often times but it doesn't get both the metro area and the high desert. It puts 9dBd into the local areas but has a doughnut hole

Out at the beach 100 miles away its solid! A 5 watt HT does just fine.
The roads and hiking trails in my immediate vicinity would benefit from a 3dBd gain antenna. In fact, the purpose of our repeater system is that of a REACT-like service for the traveling and recreational ham. We want them to be able to hit one of several machines from damn near anywhere whether or not at the bottom of a canyon or a highway out at the edge of our coverage. I travel 200 miles nearly every weekend and our repeater using a 6dBd Station Commander 620N and 50 watts out to the antenna is loud and clear - usually full quieting - at the outer edge leading into the Tehachapi Mountains. It even makes it into Tehachapi that is covered by four 440 machines. Our 120 mile coverage remains < -70dBm! (30 over-ish on RX.) We would benefit in our mountains more from voting.
But to my point about the low-level site we have mobiles and HT's (we are designed for HT's) that hit us from the flat metropolitan area of Southern California. For this reason as part 1 the side-mounted 9dBd DB224 blankets our coverage area to the outer edges. Modeling with Radio Mobile Deluxe clearly shows, and is supported by actual testing, that the 6dBd antenna does not cover the beach communities nearly as well from the low-level site. Part 2 is that this particular site that I am not naming as a courtesy to a discourteous site owner sits in the middle of several mountain areas that benefit from a signal generated right into its heart. The site is far enough away so that the pattern can spread vertically and on the DB224e the vertical beamwidth is only 16 degrees to begin with whereas the Station Commander is 70degrees. I agree that 9dBd without down tilt would be wasted RF. The 1st design goal that we not out-talk our receive has been met. Our 2nd goal of primary access via HT has not. Using Allstar we can affordably vote equipment but not sites.
On a parting note, I ran an experiment from my home. When we were putting up the 85 foot tower and DB224 we started on 146.520 and moved to another simplex channel. We had users calling "the guys putting up the repeater on the hill" from over 120 miles away. We were using 1 watt HT's to talk during construction/antenna installation!!! The later experiment was while on .52 where my 50 watt base dominates. We were talking about beamwidth and I started on a 6dBd omni and then switched to a discone. I then switched to a 9dBd omni and talked all the way up north to a ski resort (Mammoth Lakes) several hundreds miles away. People began asking who I was talking to and assumed I was faking it until I was humbled by Elmering on the topic of how gain affects our beamwidth. My distance to the horizon plus the other station up north at its elevation and the zone in the middle that was created. One ham firmly declared that it was impossible! for me to talk simplex to a station hundreds of miles away. He would be wrong as most know. The
geometric range is the square root of the antenna height x 2. In my case 112 miles. My actual service range factors in refraction. As some know many hams don't understand how we get gain by narrowing the aperture of an antenna. It is further complicated by introducing Pythagoras Theorem. Here's a calculator to which you add 20% for average refraction.
http://www.boatsafe.com/tools/horizon.htm It does not factor in topography or obstructions. Maybe this will help someone.