There's an idea of the "right play" or the "perfect play"—that is, a single correct course on any given board. Such a thing might theoretically exist, but I'd argue it only does in such an abstract way as to not be useful to strive to find. I feel like I'm really saying something here. I'm legitimately arguing against trying to find the perfect play, so let me explain.
I believe that if you and I were to open the same pack in a draft there's a good chance that the card you should take isn't the card I should take. Sometimes we open an Elspeth, Sun's Champion in Theros Limited and it's difficult to argue against taking it, but most of the time our preferences about archetypes and the different experiences we've had will inform the path that we'll take from each card we could take in this path differently. If you've drafted red twice as often as any other color and generally win when you do it and generally lose when you don't, you should probably take Lightning Strike even if I've been more successful drafting blue and would take the Voyage's End.
People want to feel like they can know that they're doing things right. After all, if you lose but you know you did everything right, it's not really your fault, and you can just feel good about having made all the right plays. I have two replies to that. The first is that you can never know this so don't waste your time worrying about it, and the second goes into a little more detail on that I suppose. Sometimes the "right play" is a losing play. Sometimes when the right play is a losing play, there is another play that would be a winning play. Can the only winning play ever really be the wrong play? This is the basic idea of doing the right thing for the wrong reasons, and in Magic it all gets very complicated very quickly.
You have a 2/2, your opponent has a 3/3, and you have no hand. Most of the time attacking is going to be strictly wrong. What if your only chance to win is to get the two damage in? It's still wrong because there's absolutely no way your opponent should ever not block when they know you can't do anything because you have no hand, right? What if your opponent has already demonstrated that they're really bad/oblivious and most likely haven't noticed that you have no hand and might give you credit anyway?
After all, I did preface by saying this is the only way you can win, so you have nothing to lose. In fact, in this case attacking is clearly the right play, but of course against an opponent this bad maybe you have other ways to win. They could always just forget to attack you for the rest of the game and deck themselves if they play enough card draw, right?
Maybe you're not interested in how best to play against terrible players, but remember that that's only a relatively slight exaggeration of every experience you have. You're always playing against a flawed opponent, so games will have more in common with playing against someone who might make mistakes that you can play to than playing against someone who always does everything right (if that's even a thing).
I guess another way to think of this is that often the correct play isn't even something that will enter into the list of plays you've considered, such as the famous bluffing an intent to block with a creature that can't block just to intimidate your opponent into not attacking. Magic is a game of hidden information and bluffing, and often the "best play" in one second might be different than the "best play" the next second because that extra second pause will change your opponent's assessment of what's going on in the game and make them react differently.
But, I've gotten lost in muddying the issue and explaining why this can be difficult without getting to how to work through it, find some clarity, and make the right play for you.
You have to know yourself as a player.
You need to know what kinds of strategies you're drawn to and what kinds of games you play best. Are you good at creature combat on a crowded board, always knowing when to attack with what, how to block, and which tricks to play around? If so, you might want to try to avoid trading, avoid racing, draft creatures with high toughness and good activated abilities, and just look to create complexity wherever possible. If you're bad at such things, you should try to draft creatures with high power and removal spells and attack early to try to trade or end the game. It's pretty easy to engineer games such that you'll consistently have few creatures or consistently have many creatures in play if you're consciously trying to do that.
Similarly, are you better at playing tempo games, where life totals are important, or are you better at playing attrition or board-control games that are all about building a sustainable dominant position? This brings up similar questions to the last paragraph, so let me go somewhere different with this. How do you know which of these you're good at? Well, if you're good at looking ahead and thinking through the math of how a race will play out and accounting for your opponent's next plays that are likely to change that math, you might be good at tempo games. If you're less good at planning for the unknown future but better at seeing all the angles with all the cards in front of you and coming up with a long-term plan, you might be better with a board-control deck.
(A realization I've just had that is interesting to me: I think tactical thinkers are likely to be good at looking ahead two or three turns and playing tempo, while strategic thinkers might actually be worse at looking two or three turns ahead, which is weird because I think of strategic thinking as being about planning ahead. But here I think a strategic thinker will be good at thinking three to ten turns ahead rather than worrying about how the specifics of the immediately following turns. All of this is very blurry, so if it doesn't resonate for you just ignore it. It's a side point.)
The fundamental point that I'm making here is that the correct play for you needs to take into account how you're going to play from that point on. To use another example I mentioned in a recent article, siding Gainsay in for the Mono-Blue Devotion mirror might be right for me, but if you're going to tap out every turn even if you have it in your hand, you might be better off drawing a different card and might not want to side it in even if you already made the decision to put it in your sideboard.
This issue is further complicated by the fact that different plays will be right against different opponents, and this is even more important. If you know your opponent just hates taking damage and will always block (common in extremely new players), you should never bluff an attack with a small creature into a large creature, but if you know that your opponent hates to feel like they got "blown out" by a trick and instead wants to try to deny you "value" on those cards and will rarely block a small creature with a large creature, you should usually just get in there. Sometimes you need to play the players as much as you need to play the game, and it's important to remember that that means playing to yourself and your own strengths, not just playing to your opponent.
I've mentioned drafting, sideboarding, and combat—I need to be clear that this is important at absolutely every stage. Part of why I think it's so important for all players to know how to build a deck is that I think it's important to build a deck you understand with cards you have a specific plan for so that you never draw a card that you just don't know what to do with.
Take a deck like Jund. Jund is a perfect example of a deck that different people are going to play very differently—it's a midrange deck that can take any role, and different people are going to think of what they're trying to accomplish with the deck differently. Some people will look at creatures and Rakdos's Return and think, "I have creatures that can attack, and I have an X spell. I should try to use my creatures to get my opponent to a low enough life total that I can burn them out with this X spell."
Others will look at the same deck and think, "I have creatures that can win the game if they go unanswered and a Mind Twist style effect in my deck. As long as I can live long enough to make my opponent discard their hand, my creatures can easily take over this game." The first player will likely cast Olivia Voldaren on turn 4 and start attacking with a flying monster. The second player will likely cast a removal spell or maybe just pass with four mana with Olivia because they want to draw out the opponent's answers with their Thragtusk and then Rakdos's Return before dropping their trump card.
Is there a "right way" to play Jund? Is one of these players wrong? Maybe, but I'd argue instead to just realize that while these players might be playing the exact same cards they're in some sense playing different decks. Obviously, literally it's the same deck in the sense that it's the same set of 60 cards, but when we talk about decks, we're often talking about strategic archetypes at the same time. Is Jund an aggro deck, a midrange deck, or a control deck? Yes! It doesn't depend as much on the cards in the deck as people think; it depends on how they're played.
If you play Jund as an aggro deck, you'll do better against decks that are bad against aggro decks, and if you play it as a control deck, you'll do better against decks that are bad against control decks—it's like you're choosing to play a different deck just by choosing to play it differently. Of course, this has the advantage that you're not locked into playing it the same way in two games in a row while you're locked into playing the same cards.
Often with Jund in particular, we'd see people play many cards in common, but the last few slots would make it clear how the person thought about the deck. This led to naming distinctions where we could call a list "Jund Aggro" because obviously they wouldn't put Dreg Mangler in their deck if they didn't think they were the beatdown. This is the great thing about knowing your deck and being able to tune it to your playstyle.
It's not unreasonable to play three maindeck Dissolves in Mono-Blue Devotion. I don't do it because I play the deck as a tapout aggro deck, which the deck I want to be playing in this metagame, but if I wanted to play an aggro-control deck where I planned to play creatures for the first one to three turns and then pass with a counterspell up or at least with the threat of a counterspell, I could certainly do that. Would it make my deck better? That depends on what I expect myself to do with it when I draw it—if it's going to trick me into passing when I need to be aggressive, it might be worse to have it in my hand than nothing. I'd like to think I'm a better player than that, but I don't technically know that I am.
If I'm drafting a Cube, I'm likely to play very differently if I draw Balance. Balance is an extremely powerful card, but it's one that I'm going to try to set up in a certain way. If I tracked results well enough, I might find that I actually lose most of the time that I draw Balance. Despite knowing that this card is great, it wouldn't surprise me too much to learn that that had happened—I overreact to the card and follow a game plan that is about it instead of about merely winning a game of Magic that I might be more likely to win by just playing my other cards.
Another more immediate example: Marshall Sutcliffe recently asked about cutting Library of Alexandria from his Cube deck:
Do I even bother with Library in a deck like this? http://t.co/zY0YgZkvZT— Marshall Sutcliffe (@Marshall_LR) December 31, 2013
Before looking at the deck, I thought, "Of course you do not cut Library; this is absurd. He probably drafted some aggro deck, but that doesn't matter. He should definitely be playing Library." I genuinely believe that his deck would have been objectively stronger with Library of Alexandria basically no matter what other cards in the Holiday Cube were in it, and the aggressive red deck he had definitely wasn't a counterexample to me.
When you draw Library of Alexandria, you're not obligated to play it on turn 1, activate it, and hold seven cards in your hand throughout the entire game to keep it going—often it's right to do that, but it isn't always right. Despite knowing that the opportunity cost of the option to use Library of Alexandria is certainly low enough that he should play the card in theory, assuming he will always correctly identify when to use it there is still a case to be made that he's more likely to win without it.
His claim was that playing and activating Library on turn 1 is worse than playing a two-drop because he's just trying to race his opponent, and his cards are going to be low impact in the late game so having them then won't help—often a 2/1 on turn 1 is worth more than four 2/1s on turn 4. That's valid, and if I were to play his deck, I wouldn't be at all surprised if one eventually concluded that I was actually losing more games where I drew Library of Alexandria than games where I didn't simply because I couldn't get away from the allure of using such a powerful card and erred toward doing it when I shouldn't have.
I'd want to play it on turn 1 against an unknown opponent. If that opponent was playing reactive answers, that'd be great, but if that opponent was playing a dedicated storm deck that I had to just race, I would likely be hurting my chances by not taking the most aggressive line. The fact that the card gives me the option to use it to beat counterspell decks means that it should be in my deck, but the fact that I might not be good enough to avoid losing because it tricks me might outweigh that in making a case against. Of course, when someone asks what the correct choice is on Twitter, I'm going to assume they're asking for my closest approximation of the theoretically optimal play assuming an omniscient player.
If you can only win games of Theros Limited with fifteen-land mono-black decks, please take Tormented Hero over Elspeth, Sun's Champion if you need to win that draft. Just because it's "wrong" doesn't mean it's not the play you should make.
(This is also why Magic is beautiful [beautiful in that it is an art rather than a science].)
(Reason #279, if you were wondering.)