Not knowing what else could be in my opponent's deck, I'm reluctant to loot because I don't want to be in a situation where I have no answers left in my deck for another threat of a similar caliber. However, I think what LSV says about looting is correct, just not necessarily for this specific example.
A big part of this is that we're substantially advantaged (empty board, we have a looter) if we just untap and Lightning Blast the angel. So we have a situation where the most important question is, "What do we lose to?" And besides the idea of our opponent answering the Control Magic somehow there's the more probable one where they just have another threat that doesn't die to Shock.
If you loot EOT, there's around a 4% chance that the card you see is Control Magic, and you discard the Lightning Blast, and you're now losing to a bigger flyer, or disadvantaged to a removal spell followed by a 3/3. So it's not a disaster but it's also one of the only ways you lose this game. This wouldn't be nearly as bad if we had another reasonable removal spell left in our deck.
The upside is that you see one more card. In which situations is seeing that card the difference between winning and losing? It's basically only those where the extra card you see is Control Magic, and it's the extra card you see because your opponent had an extra threat that you otherwise die to. So assuming the opponent has a threat (which we might as well because we already win all the other games), that's the same 4% as above. There's also the case where you would have seen the Control Magic in time without looting, but the loot allows you to Control Magic on 5 life instead of 1, for example, which does affect your chances of victory somewhat. If the opponent could have burn in their deck I would assign this some extra value and that could push us more towards looting. Other than burn, what we lose to from having stolen their backup threat is very similar regardless of whether we're on 1 or 5 life, so I don't think this narrow case adds value to looting.
So if we're 4% to make ourselves vulnerable and 4% to save ourselves from a vulnerable position, it seems like it's essentially neutral. What other factors might influence our choice? What other consequences would looting have? The card we see is 4% to be Control Magic, which is kinda bad as we accounted for above, 54% to be a spell which we immediately discard, making our deck marginally worse, and 42% to be a land which we immediately discard, making our deck marginally better. It seems like looting in this exact position is very slightly wrong, but it's very close.
But what LSV really meant to say, as he puts in his answer, is "you have your second-best card in hand and your deck contains a much better card and a bunch of generic mediocre win conditions" and given that, do you loot?
And it's hard to answer this in a vacuum because Magic is complicated and there are so many edge cases. Every little bit of information pushes you one way or another. But I've thought about this problem for some time now and I've come up with an example that I believe illustrates LSV's point:
Your opponent has no creatures in play and no cards in hand. You have three Islands and three Mountains, all tapped, and an untapped
Merfolk Looter. You have a
Moon Heron in your hand. Both players have 16 life. Your deck contains:
Hypothetical Situation
Here, I think looting EOT is unambiguously correct. You would absolutely always like to draw
Air Elemental and would happily throw away
Moon Heron for it. So you're 4% to be totally happy with the choice immediately. You'd discard any non-
Air Elemental card you drew, but you'd be more likely to see the
Air Elemental on any following turn in that case. Note that unlike an answer, where drawing it earlier matters only if the opponent has something at that moment, drawing a threat earlier improves your clock and decreases the number of cards the opponent gets to see. So supposing the game goes on for 7 more turns, that's 6 more positions in the deck the
Air Elemental could be where you draw it earlier because you looted and it gets to attack, for another ~23% that looting EOT helps you win faster.
(7 turns is how long it will take to kill them with
Moon Heron, but you could easily draw other threats that would shorten the clock, or the opponent could be able to interact. It's beyond the scope of this model to examine the exact consequences of each series of draws.)
So instead of looting EOT having 4% of a downside and 4% of an upside as in LSV's example, we have 4% of immediate upside and 23% of delayed upside. So in this case we definitely loot.
What's the actual distinguishing factor here? I think in his example, if we had another flexible removal spell in our deck, the downside is diminished and we should loot. In my example, if we didn't have any win conditions besides
Moon Heron and
Air Elemental, there would be a substantial downside to discarding
Moon Heron and looting would be less good and possibly wrong. After all, once our deck has only one threat in it, we lose to one removal spell. A 4% of losing the game outright to any removal spell probably outweighs a 23% of drawing the other threat one turn earlier.
So, in my example, there's no likely trajectory the game could take where you need to be able to cast the
Moon Heron specifically in order to win the game, because there are other cards that fulfill that role that you could draw. In LSV's example, there
are trajectories where you need to be able to cast the
Lightning Blast specifically in order to win, because you have so few outs to a second threat.
Both examples live in the same region of the game space, where we are advantaged and so we're looking to minimize our opponent's outs. If we were behind, the logic would change. If in LSV's example the opponent had a second
Serra Angel already in play, I think looting is definitely correct because even if we
Lightning Blast one of them we're on a strict clock to draw the
Control Magic, and the
Control Magic creates board parity by itself, and we're in no position to assume the opponent will have anything else because we can't beat any of it if they do.
I also think LSV's logic is correct in the majority of real game situations. There are occasionally positions where you're in more of a gray area where there is ambiguity in whether looting is correct or not and you need to examine the entire game state, but that's far from a common occurrence. "If you shortcut to 'loot any time you have anything but your best card,' you will be right
way more often than not," as he says. And no one has time to write out a full analysis in the middle of a game. You just take the shortcut, because brain resource management is a thing, and other decisions will affect the result of the game more, so you should devote more thinking time to those.
