Interesting advice on creating new material. Honestly, a lot of content from my blog uses ideas from existing games and media, but here's a secret that I probably don't make explicit enough: for the sake of communicating with people, I use names and concepts from popular media, but I absolutely never play with any of the things I have made up on this blog without a bit of personalized dressing.
One of the cool things about table top games is that they're somewhat imaginary. I could probably stat out any popular character, ability, or what have you in GURPS from one of my favorite books, movies, anime, or video games. Thanks to the concept of tropes however, there are a ton of ideas that have some overlap and some originality, and some inspiration. So, I play a few video games with an annoying tiny fluttering fairy enemy, and I came up with two different articles on two different approaches because I think, "That's a concept that is interesting that I have not seen in GURPS." Now I have a bunch of mechanics, and if we abstract away the game that the idea came from, and give it a new name, or describe it as looking completely different, give it a different lore or historical background, it's a new thing, and it doesn't look corny when you have a "pokémon" which is now some kinda megafauna next to some "Kingdom Hearts enemy" which is now some kinda fairy.
Different strokes for different folks. Some people like hosting games where you are inserting your character into your favorite popular franchise. I think I might be a bit embarrassed to do that. For me though, I hear people say things like, "GURPS can't do System x", or "What system is best for setting Y? Please don't say GURPS." and it kinda inspires me in a weird way to say, "yeah, that setting is so easy to do in GURPS," so I crunch out numbers and do that, and in the process, I learn something new, and now I have a seed for a new idea.
Though, going back to the original article, I agree with all the mechanical concepts. Trying to straight up take another RPG that has much different assumptions and calculations makes a "conversion algorithm" a ridiculous prospect. I prefer taking a concept I like and modeling it around what I need it to be.
Wow, what an incoherent ramble.