I've done more experimenting with sound ports than I ever meant to over the past several years. These include building three 'test mules' with various port configurations, making reams of measurements such as spectrum charts and resonance plots, and a 'bind' listening test that involved more than 100 players over a couple of years. Here's what I found.
Ports work, at least for the player.
At low frequencies the sound of the guitar goes out in all directions; it's a 'point source' for the fundamentals of the lowest notes. As you get up around the pitch of the open G string it starts to become more directional, and at high frequencies most of the sound is coming off the top and out of the hole, going toward the audience.
You, as the player, only hear those high frequencies if they get reflected back toward you. Having played once in a large, dead space with a sound system that had no monitor, I can tell you that it's disconcerting (note pun), particularly if it's also cool enough that you lose sensation in your finger tips. A port 'hears' some of those high frequencies, and one that you can look into will send some of that sound out toward you. It doesn't have to be very big to work pretty well: your hearing tends to get more sensitive at high frequencies, so you don't need much to fill in the blank.
Naturally, as an added sound hole in the box, the port can change the timbre of the guitar. Any port will raise the pitch of the 'main air' resonant mode and make it a bit more powerful. how much it changes depends on how big the port is, and where it is: basically, the larger the port and the further from the main sound hole, the greater the change. Usually the pitch change has more of an effect on the timbre than the added power, but where you end up depends a lot on where you started out. Because there are so many variables it's hard to predict exactly what a port will do in a given guitar.
The listening test I did used a rough, but decent sounding test mule guitar with a 52mm diameter port in the wide part of the upper bout facing the player. This is larger than I often use, but is fairly common. The port has a 'door' that can be removed and replaced quietly. I had people sit with a blindfold on, and handed them the guitar. Once they had played it, I took it back for a couple of seconds and had them play it again. I then asked them if it sounded 'the same' or 'different'. Each time I handed it to them I used a coin toss to decide whether the port would be open or closed. About 50% of the time the configuration was 'the same'. The tests were run at two successive Montreal Festivals, a GAL convention, and a local folk festival. The average noise level was about 70 dB-A.
When the port configuration was unchanged, the players guessed, getting it right about 50% of the time. When it was different, they got it right almost every time. The people listening were not polled, since for them it was not a 'blind' test, but many of them volunteered that they really could not hear any difference when the port configuration was changed.
Other testing, done by an associate, using a different test mule I loaned him in quiet, reverberant spaces, showed that players could not hear any difference when the configuration was changed.
So, the bottom line is that a port might well be useful to you if you tend to play a lot in large, 'dead', or noisy places (a 'restaurant gig'). Otherwise it might not help much. Some feedback I've gotten indicates that people with hearing loss often find a port helpful. That makes sense, since the usual hearing loss involves decreasing high frequency acuity.
I could fluff this out with a lot more detail, but that's the gist.