Quote from RMelanson »Re: "Isn't it contradictory to say that you can't find the mana to suspend Ancestral when Search costs two?"
The answer is yes and no. It's more a question of using the mana efficiently and optimally. Breaking down the curve of the deck as it currently stands, there are a lot of 1 drops, few two drops (Terminate, Logic Knot, and Azcanta), and then lots of three and four drops (including Snap-one drop; Snap-two drop). (There is also the consideration of dropping T2 Tasigur, which suspend AV makes more difficult, but not impossible.) The tl;dr is that the deck has a lot of options for Turn 1 play, that gets pushed to Turn 2 if you suspend Visions.
For example, lets say you're on the play, drop land, AV. Then on turn 2, you usually are using only 1 of your two mana, maybe casting Terminate, or just holding up mana for a 2 mana Counterspell. In this scenario, there is a substantial risk of efficiency loss. In contrast, if you react on turn 1 with either a thoughtscour or removal spell, and then hit azcanta on turn 2, then you are using all your mana efficiently. Over the next three turns, you have Azcanta action 3 times, increasing your chance of drawing relevant action. You also have the ability to ramp up mana once you hit 7 in the GY, which can help you use more spells to protect Jace. With AV, you draw additional 3 cards on turn 5, that can vary from very good to lands/fluff. On the draw, both AV and Azcanta can be clunky, but Azcanta at least offers value sooner when you eventually cast it. If you are facing off against an aggro deck for example on the draw, you can T1 hold up Spell Snare or use removal and then drop Azcanta when you have time rather than feeling pot-committed to dropping AV.
This analysis obviously changes based on the meta and how your construct your deck. Imagine a world where BBE is king and the field is just Jund. In that world, at least you can drop AV on the play under your opponents discard. With Azcanta, you just need to pray that your opponent finds Snapcaster or Jace more appealing to take. Or imagine we are in a world of blue decks, AV is nice because on turn 5, you get to ask your opponent, "hey, are you going to counter my Draw 3, or this Jace I will likely play otherwise?" You can also avoid a lot of the choices I mention above by just building different. You could take out Spell Snare to avoid the conflict with AV. You can put in more sweepers to compensate for a slower response and the likely uptick in Lingering Souls. If there is a clear answer (which there probably isn't), its not between choosing bad or good, but more like great or marginally greater (black v. slightly darker black).
The mana efficiency issue you pointed out isn't nearly so simple. In your specific examples sure it works better along the curve but for every one of those there are tons of others supporting ancestral over search. One mana will always cost less than 2, it allows you to start casting multiple spells in a turn earlier and especially late game with a flipped search. This comes up a lot in control mirrors where you or your opponent tap 4 to search and the opponent responds by trying to resolve a high impact spell (I've run into several from secure the wastes/rev to breach in ur combo/control or against jeskai they try to burn you out in that window). Having to bottom snapcasters with search also feels really bad.
Don't get me wrong search is a great card I'm a big fan of it but it's certainly better in matchups where you value the immediate impact of the "scry" over long term games against other midrange/control decks that have more discard, counters or land and enchantment hate. I think it's a meta dependent call and if the meta does slow down I see ancestral being better.
Also Tiemuuu crazy how quickly things change. A very short while ago I would have been very strongly against cutting any number of scours cause it just synergizes so well with the deck. With less delve creatures and a high impact card like Jace you don't wanna mill over I think it does lose value though. Still a good card but not sure if it's necessary if we're building in that direction.