I think discover should choose 2 cards instead of three. For example, just a bit unfair, a 1 cost lackey discover a dragon or spell and swing the game.
Personally I like discover, I just don't like discover being limited to such a small range of things and always pulling out the best 1 or two cards in a situation.
I think limiting to 2 cards won't really solve any issue apart from making discover just a bit worse and still if they get the spell or dragon that is still good out of the two cards you will be salty.
Discover has always been 3 and it's good as is. Nothing changed, consistency and tempo beat discover classes. Played hunter and warlock this month, with no discover cards other than zephyr, which is anyway 'cast for lethal' so it could be one card.
I think that if a dragon lackey gets you, with the dragons' curve being above 5-cost, your deck runs out of fuel a bit too fast.
As is, 58% of the time you get a card in the top 25% of potential cards - so 42% of the time, you are getting a card that is, at best, isn't even in the top quarter. 42% is almost half the time. If we widen it to 50% you're very likely to get a card in the 'top half' of cards (87.5% chance), but that means 1 time in every 8 (on average) you discover a card, you're going to get three duds that aren't even in the top half of cards.
The 75% figure is important because in constructed things are pretty cutthroat - and I'm pretty sure not even the full top 25% of cards would feature in decks, many not making the cut for various reasons (potentially not strong enough in the face of stronger options). The 50% figure is important because a 50-75% card might just be 'ok' (think Chillwind Yeti), but anything sub-50% is likely to be trash.
Long story short, these probabilities aren't high enough that discovering is a problem. Now, the specific pool that can be discovered from - that might be an issue. For example, despite these figures I have no issue running Dragon's Hoard in my Galakrond rogue deck. You're likely to hit a top half or top quarter legendary minion, which should at least end up being serviceable - but because of its ability to give you off-class legends, its potential to high-roll and snatch wins is arguably worth a card that might just end up getting you something on the power level of a Yeti or Boulderfist Ogre.
I'm fine with the Discover mechanic, but it should be used FAR less frequently then it currently is. And I disagree with what someone stated earlier about the size of the Discover pool. I think the Discover pools should be limited to only a few cards and NOT a huge pool of options with such varying strengths as the difference between discovering Fairy Dragon and DQ Alex is game winning/losing after playing a 20 minute game.
So they should limit/control the Discover card pools better and create less Discover cards. Then the mechanic will be fine.
here comes Valeera and half a deck worth of legendaries created by lackeys. Doesn’t matter that the deck is full of trash and no late game when you can just discover every card in the game through steal mechanics and discover over and over via lackeys.
Discover mechanic adds quite a bit of RNG, but also quite a bit of skill at the same time. It's not very common that the choice among the 3 offered is straightforward and obvious and quite often you need to decide whether you will make a long-term pick that might and up getting stuck in your hand or make you the king of foresight