Happy to hear you enjoy it!
Ok, for the technical bit, let's define some terms. A word is any word, reused or not. "word hello word" has 3 words. A unique word would be any word featured in the article one or more times. "word hello word" has 2 unique words ("word" and "hello"). A solo word is a word that exist only once. "word hello word" has 1 solo word ("hello").
After I had published the game (I could only work on Sunday, so I was crazy rushed), I added a debug feature to my game to get some statistics. I opened maybe 10 random articles, and more or less no matter the length of the article, roughly 2/3 of the unique words were solo words. However, only about 15% of all the words were solo words. That means that about 85% of the text is comprised of 1/3 of the unique words. Fascinating! Anyway, the fact that it was as consistent as it was surprised me. After having played around a lot just guessing in my browser with ctrl+f (my prototype!), I felt quite confident that I could pick any article and it would work, but that a longer article would be more fun. I then just went to the page where Wikipedia puts featured articles and put them all in the game, so those are the 15 articles of which you will receive one at random. I actually had only tried one of them when I uploaded the game, which made me slightly nervous :D The reason I wanted to get some statistics afterwards was that I got lucky a few times when trying it out and felt it was too easy. Then it turned out that all the articles I checked had hundreds of solo words each! That became even more discouraging when I compared it to the number of unique words to find that 2/3 unique words were solo words. But then when I realized I had to compare to total words and found about 15% I was satisfied at last. But there are about 6 million articles on Wikipedia, so there might still be one or two without solo words, although I think humans also actively try to vary their language, which helps if you're a solo word hunter.
On the more computer technical side there's especially one technical thing I found surprising. Overall the game is very simple - the tricky bit was mostly to make it really feel like you're browsing Wikipedia while being able to check if the word is actually a solo word, because browsers have a LOT of security against reading any type of data from any domain other than the one you're currently on, so as soon as I show something from Wikipedia, I can't access it at all from the code if it even loads in the first place. I still didn't want to just download and republish copies of the articles though, because it just felt wrong and would probably be illegal and difficult to pull off well either way. In the end I had to use the official Wikipedia API with some special cross-browser settings on both ends, extract the data from there and use the extracted data to compare against the words you enter. At the same time I load an iframe with the same article so you can scroll around and read. In other words, the article is actually downloaded twice in different ways, and what you see from Wikipedia and what goes on underneath the hood are completely separate from each other (which is also why you can't guess a word by double clicking it).
This became a bit long, but that's what you get for asking fun questions :) I'll gladly tell you more if you're interested.