Sorry but wrong. "Blind" tests are in reality rarely done because a) a good scientist will have some idea of the provenance of the rock sample and b) it costs a lot of money to keep trying different dating methods, hence a). But that isn't to say that you couldn't go on a blind date if you wanted to. If you were really interested in finding the age of a sample with no prior information, you do it with several methods and you do them on individual minerals separated from the rocks. Then if the methods agree you almost certainly have a good date.

In the real world, of course, you essentially always do have prior information.

The K-Ar method (and almost all radiometric methods) boils down to measuring a quantity of stuff. (The best methods actually measure the ratio between the quantities of two different stuffs, and then use other information about the quantity of one to calculate the quantity of the other.) When a rock is old, there's a lot of stuff to measure. When the rock is young, there's not much stuff to measure. A K-Ar lab will start by cleaning as much previous sample material and argon from the equipment as is feasible. (If you tell them you think your sample is relatively young, they will take extra care in the cleaning process. And charge more.) Then they will measure your sample and the readouts will tell them there's Y amount of argon.

Every few runs they will run the equipment with nothing in the sample holder. The readouts will say there's a non-zero amount of argon, call it X, which they can calculate as being equivalent to an age of Z years.

If the amount they measure in your sample, X, is pretty much equal to Y, then they will report the age as Z years or less. If X is greater than Y they will report an age older than Z with appropriate error bars. If X is less than Y they will try to figure out what's going on. Add to this the fact that the K-Ar method is simply unsuitable for very young rocks. In essence, if you submitted a sample of young rock for K-Ar dating, you would get a result back that says something like <100,000 years, in which case you try a different method more suitable for young rocks.

I find myself having to repeat the same thing as always: don't criticise things you don't understand. Don't you think that's a sensible rule of thumb? Like, cherry-picking articles from scientific journals on nuclear physics for words like "inaccurate" and "discrepancy", when the people who wrote those articles and those who reviewed them for the journals know a little more about the topic than you?