I don't think so. I believe that any irrelevant parameter increase the instability. We all actually observe such instabilities and most probably because of the irrelevant variables.
Here is simple example how it may happen. Let us imagine that someone decided that the word "hat" might be important for the fashion sites. It inserts the word “hat” and trains the robot, giving it the good and bad sites as examples. If it is the time when hats just become to be in fashion, there will be high correlation between the good sites that noticed this trend and the word “hat”. The word hat will get high weight because of the training.
Now what will happen when the hats becomes out of fashion? Probably, really good sites that noticed this trend first will loose their high index position because they stop using the word “hat”.
So irrelevant variables increase the instability and require manual corrections in data mills also.
Bottom line. If a parameter has a little relation to the good content, like, for example, the period of the domain registration, do not bother about it.
Even if the Google will use this parameter, it stops to do it quickly.
If it is irrelevant it is irrelevant.