According to PC World, AT&T's 3G service improved in Boston by 184% (nearly doubling) between Feb 2009 and Feb 2010.
Public perception is one of those things that doesn't change quickly (no matter what the evidence). I know that a lot of people had billing issues after the Sprint-Nextel merger and that's been resolved and Sprint's service is as strong as I've seen it, but customers still have a bad impression of Sprint.
With wireless service, it's hard to really get good data. Verizon is trying to get people's opinions to be high of them based off of their map ads. However, what you really wants is strong coverage where you are rather than weak yet broad coverage in places you aren't. Coverage isn't a binary situation, but mapping broadness of coverage seems to be what people have latched on to. It would be great to see signal strength and speed measurements on a street level of all the different carriers. It would be great if Google could hook up 4 cell phones to their street view cars to measure that (if they aren't already) and then map the signal strength on Google Maps. I'm sure the carriers would object to such neutral data since they'd rather sling ads at each other, but consumers make better purchases when they have more objective data.
Speed is also harder to quantify since you likely don't notice it without the use of tools. It's easy to notice the binary condition of "AT&T doesn't have 3G in rural Maine", but harder to notice, "AT&T is 39% faster than Verizon in Boston". For what it's worth, Boston seems to be one of AT&T's weaker markets (with Sprint taking top honors in PC Mag's test; yay!) and only beating Verizon and T-Mobile in the 40-50% range.
The problem is that people want to believe that one option is better. There is one wireless carrier that if I always stay with them will give me the best service. There is one brand of car that will always have better engineering. There is one phone company that will always produce a superior product. A lot of the time, product lifecycles make a huge difference. People like being consistent - think how Kerry was labeled a flip-flopper. There's a huge social cost to saying that now you think something else is a better option and almost no acknowledgement that someone could be right *both times* even as they're recommending different things at different times. People don't accept that change happens.
Think of the iPhone 3GS. When it came out, it was faster, had a better display, etc. than the Palm Pre and later the Hero/Droid Eris. Then the Nexus One/Incredible/EVO came out and they had a higher-res display and a faster processor. Then Apple comes back with the iPhone 4 which has an even better display, better form factor, etc. Companies are often leapfrogging each other, but people want to believe that they're always using the best so they justify, ignore evidence, and even downright lie to make themselves feel like they're never on #2. I mean, Sprint has 4G right now, but I don't expect Sprint to always be in front of everyone just because they're the first to 4G - life is more complicated than that and companies change position in an industry a lot.
As a Sprint customer, I'm more than happy with the service I get and really can't get into the "my network is better than your's" that Verizon and AT&T customers spout. Verizon customers are especially bad and will usually recite ads more than evidence. It's why I find evidence so important. In this case, if someone has a less-good wireless carrier, who cares. I mean, really. However, evidence is what brought medicine to where it is today and non-evidence-based medicine is often dangerous. Yet, millions of people commit themselves to opinions and treatments with no basis in science and evidence. That's dangerous! Here, whatever: people like getting into silly arguments about things that don't matter. However, it's a bad habit to get into to rely more on anecdotes and feelings than evidence because there are places where it really does matter.