When I discovered Namecoin, it seemed like a perfect thing to use to replace the ICANN domain name system. I still think that some blockchain is going to be required to replace DNS, but over time, I've come to know that there were several fatal issues with the implementation of Namecoin. These issues have joined to cause Namecoin to be beaten by the network effect.
The first and most glaring issue is the lack of a built-in DNS protocol server for Namecoin. The software should have shipped with an implementation by default, even if it were very rudimentary. Even now, seven years after it was released, there is still no built-in DNS server for Namecoin. There exists an external program to perform this function and there have been scripts to generate a BIND-compatible zone file, but this is a terrible way to go about it. People are going to have enough trouble setting up the software; fewer people are going to be able to make it the advertised function, an uncensorable, distributed version of the domain name system, work at all.
This leads me to the next point. The project should have setup and maintained several public domain name servers for Namecoin and requested the .bit TLD from ICANN. I know that it was intended to replace the ICANN system and that it is very likely that ICANN would never allow a .bit TLD into its system, but that's impossible if there is no traction to make the change. Even if the .bit TLD could not be secure, the domains could have been exposed as subdomains under the namecoin.org domain. Having the public be able to use namecoin-registered domain names without having to install any software on their computers makes the user base much larger. It would never be as secure as running the software on your own hardware, but the issues preventing Namecoin adoption has been with bootstrapping and not with security.
Similar to the last point, there is no functionality built into Namecoin for transferring existing domain names to the blockchain. This means that to get to a world with everyone using namecoin, you have to either add this function or have everybody abandon every existing domain name, which is never going to happen. Transferring should have been included at release or added early.
Finally, the economics of Namecoin are best described as dysfunctional. The main drive to provide value to NMC, the Namecoin currency, is its ability to register domain names. Having been crippled by the previous issues, there is little reason to register one of these domains, which filters to a low utility for NMC, which is reflected in the coin's price (1.16USD/NMC at the time of writing).
All told, there was to far to strong a headwind for Namecoin to gain traction. I certainly think there is a need for the type of technology that Namecoin represents. Others think the same, which is why there have been several attempts to replicate this domain name registration function, but these alternatives have succumbed to the same issues that Namecoin has.