If it is indeed better, which would be counter-intuitive, you should probably not have any problem digging up evidence that it is so? And by evidence, I mean metrics, measurements and stuff like that. You know, scientific standard and all.
If I have DNS server 10ms away, you can bet your ass (and house) I can expect a smaller latency than from one hosted 6000 kilometres away across an Atlantic ocean and then some.
And for me lower latency means faster, and faster means better. And being on the same network as me, I can expect much higher reliability than something halfway across the internet, no matter who hosts it. To me higher reliability is also better.
What definition are you using, and where is your data to prove it?
Unless your ISP's DNS server is a 486 collecting dust next to the janitors toilet, chances are you better served with that than doing continental-hopping network-traffic for every domain you need resolved.
You should look into "anycast". The Google DNS servers are located all across the globe. For example, I get 60ms from Portugal (just +30ms than I get to my ISP's server). And Google's servers, due to higher use, are much more likely to have the records in cache, reducing the overall latency when compared to a server than needs to issue a recursive query for every less commonly used domains.
Ah, if only my ISP's DNS server were 10ms away, and reliably up, and not serving ads instead of NXDOMAIN results.
Not that I'm deeply trusting of 8.8.8.8, but it definitely wins on each of those points. (I use a different "public" server.)
Google's currently don't return false results, return fast results, and have a sane caching system. Recently when a site's DNS went down for a day, google continued serving the results, although expired. Most other providers cleared the cache first, and were serving no result.
I've setup my own caching DNS before, it's really not hard... Most home routers will do this today. And my ISPs server isn't 10ms away, but my router is.
How about: because it's better?