I really think with Mysterious Challenger around, complaining about Mad Scientist seems pretty crazy.
Well, Challenger is also a problem for the same reason.
The issue with Scientist is that for it to not be ridiculously OP, secrets have to be bad cards, or at least highly situational. After all, you're spending 2 mana and 1 card to get a 2/2 body (1.5 mana worth of tempo, 1 card) and a secret (2 mana for Hunter, 3 mana for Mage; 1 card), so you're +1.5 or +2.5 mana tempo and +1 card above par. This is partially balanced by most secrets being poor cards individually, which both makes Scientist's deathrattle effect weaker than it seems at first glance _and_ more subtly requires you to run bad cards in your deck.
(There are many exceptions where secrets don't quite fit this simplistic mold. Ice Barrier and Ice Block were used in Freeze Mage before Mad Scientist, after all. As another example, Duplicate is a secret that is good in the right deck but you might not want to draw off Scientist's deathrattle.)
Problem is, if you make cards like Mad Scientist and Mysterious Challenger, you've locked yourself into a design where secrets are bad cards that you only put in your deck so that you can hope not to draw them directly and instead put them directly into play via Scientist/Challenger. Every future secret designed has to be sufficiently bad so as to not make Scientist/Challenger more OP, but still good enough that you can justify putting it in your deck at all. That distorts the design of secrets in an ugly way.