One thing that the losers have in common is that their execution was not good enough (or outright bad in the case of ExpertsExchange).
But Altavista was huge when Google started. AOL had network effects that did not help them, so it is absolutely not clear that it is impossible.
People however are more lethargic and programmed by their mobile devices these days, so it may be more difficult now.
Google had the entire internet inedexed and was running their engine from Standford with just a few machines.
The internet was tiny back in 1997.
In 1997 there was 1 million web sites.
Today there are 2 billion.
[1] https://www.internetlivestats.com/total-number-of-websites/
If instead you want to have some fun and maybe get more money back than you put in, that's a different world.
I'm increasingly persuaded that unless you go to dinner parties with say Sequoia capital people, it's better to start small, get the principles working right and then build on it.
And Google Talk/Allo/Jibe is Jabber and Asterisk with a fancy coat of paint.
There is a net demand for Chinese food in the area, and there will definitely be such a thing as 'too much or little' supply (though obviously each offer i.e. supply is a little different)
A single restaurant will only be able to meet so much demand, so if there's a niche in terms of region or menu, maybe there's a real opportunity.
GitHub, Uber, FB - they all have network externalizations to some extent, there definitely have economies of scale, and they are playing in a global marketplace (or at least, say 'the Western world').
So, no, it's probably not a good idea to compete directly with them.
In fact, incumbents almost always win - the idea is to do something different enough, solve a problem in a different way.
Slack did not build 'better email' they went up one level and built 'better communication'. And a whole bunch of other things of course. But 'slack e-mail' probably would not have yielded them the same result.