More commonly linguists talk about Swadesh lists. For the most complete modern collection of Swadesh-like lists and a program/algorithm for automatically generating a language family tree from them see: http://asjp.clld.org
While ASJP is used to explore potential relationships, it should be noted that it cannot reliably determine language relatedness by itself. The gold standard for that is still the comparative method. [1]
"You" was plural and/or formal, "thou" was the singular/informal form and so, per the list, "thou" was item #3 of the list of "words least likely to be replaced as the language evolves".
The research behind that article is extremely problematic. See e.g. [1] and [2]. The WP article adds its own overinterpretations to it, such as as the nonsensical "But if you went back 15,000 years and spoke these words to hunter-gatherers in Asia in any one of hundreds of modern languages, there is a chance they would understand at least some of what you were saying."
That's easy - until very, very recent human history, the louse/lice/nit has been a constant (and really annoying) companion of much of humanity. Not surprising the word stabilized in populations that used it all the time.
What's interesting - is you could could probably use "semantic stability" to track preponderance of various pests, diseases, etc... in various populations/locations.
