So there was a bug in OpenSSL. Big bug, yes, but that's not the reason it was (and still is!) a big problem.
The genesis of the big problem is one of monoculture, not only of OpenSSL being the dominant SSL implementation, but probably more importantly, the fact that pretty much all Internet security that is accessible and matters to ordinary users is SSL/TLS in the first place.
If you think this is bad, imagine what happens if the fundamantals of SSL itself are compromised: What would we replace it with? How, considering this is effectively the only secure connection technology available across all common OSes and embedded devices? How long would that take? (Years, at least, I'd wager...)
What we need is more flexible security methods in the first place, and open, standard implementations (like OpenSSL, but growable) that can allow us to proactively extend security methods as the net matures, and *quickly* address bug-based vulnerabilities when that approach fails. (Note that this may require the implementation of some kind of standard "secuirity code VM", so new code and new methods can be easily distributed even to older systems that may not be fully supported anymore. And no, I'm not glossing over things like limits on code space, memory, and the like, nothing will allow every system to be upgraded, but we do need some way to allow and authenticate that (while preventing bad guys, including governments, from using the mechanism to create weaknesses.))