There is a wealth of information out there regarding the correct way to cast Brainstorm (if that is even possible), the sequencing concerning Ponder, and why Preordain is worse then Ponder in circumstances X,Y and Z. What there isn’t a ton of information on is their inferior little brothers Serum Visions, Thought Scour, and Sleight of Hand.
Most cantrips have a minor effect on the game while also drawing a card. So technically cards like Remand, Electrolyze, or even Cryptic Command might fall under this classification. These cards are fundamentally different from the cards I will focus on though, in that they require more of a mana commitment. I will focus on the first couple of turns today and therefore will restrain myself to the 1-mana cantrips:
Deckbuilding with high cantrip numbers is a given in Brainstorm’s home format of Legacy, so taking a page out of tried-and-true Legacy mana bases might be the way to go, especially since the fetch/dual mana base that defines that format is also the norm in Modern, if slightly more shocking. There is, however, a fundamental difference between the Legacy-only legal cantrips (Brainstorm, Ponder, and Preordain) and those used in Modern. The Modern versions don’t let you control which cards you draw nearly as well. So while the Legacy versions actively help “fix” your mana, the Modern counterparts mostly just condense the deck.
Say you need to make a land drop—Ponder lets you look at the top 3 cards and, if there is a land, draw it. If not, shuffle and get another crack at it. Serum Visions on the other hand draws a random card and lets you draw a land next turn if there is one in the top three cards. Thought Scour just draws a random card from your deck.
When I say these cards “condense” the deck, I mean that for the cost of some tempo these cards let you increase the velocity of your deck—you effectively have more access to the other cards in your deck, be they lands or spells, but not one or the other specifically. Velocity is a large part of what makes Delver or Storm, the upper tier decks with high numbers of cantrips, tick. When most cards in your deck have a low impact on the game you need to have access to significantly more of them to win the game. The ability to go through your deck without breaking stride and having more access to the cards in your deck is the appeal of high-cantrip-number deck building.
With this in mind the easiest, if somewhat crude, way to look at building a mana base for a cantrip-heavy deck is to disregard the cantrips entirely. So if you play say 20 cantrips you have a 40-card deck, like in Limited. So the expected land count would range from 16 to 18 depending on the curve. You can see this approach in decks like Storm or UR Delver.
So why not just stuff all your decks chock full of cantrips? Drawing fewer lands in the late game and digging toward your good cards does sound appealing. But there is a cost. I mentioned earlier the concept of tempo. Casting a cantrip entails a loss of tempo. Imagine your opponent playing a curve of 1, 2, 3, 4, while you play cantrips on turns 1 and 2 and start committing to the board on turn 3 only. You could be too far behind to come back to an even board state even though you might have slightly better cards by now. To compensate for this, cantrip-heavy decks need to have a very low curve so they can both cast a cantrip and add to the board in a meaningful way early in the game.
If you can play a Young Pyromancer, a Gitaxian Probe and a Serum Visions on turn three while your opponent plays a Kitchen Finks, the triggered ability of the Pyromancer offsets the fact that you only spend two mana on your board commitment while your opponent spends three.
Take a look at my current build of UR Delver (which is not dead despite losing some of its late-game power with the delve spells):
Most of the deck is built to be tempo-positive, to “take back” the tempo lost through cantripping. To make this worth it, it plays mostly cards that profit from having a high spell density—Delver of Secrets, Snapcaster Mage, Young Pyromancer, and Monastery Swiftspear.
This build showcases the advantages of high cantrip numbers: Low numbers of high-variance cards (like Spell Pierce, Forked Bolt, Pillar of Flame, Vapor Snag, or Cryptic Command) get more value as you have them when they are good more often, while not drawing so many when they are not, due to only playing one of each. A low land count lets you have more spells during the course of a game than your opponent, letting you generate a sort of hidden card advantage, even without access to Treasure Cruise.
So why not play the maximum number of cantrips? Even my list only plays 13 out of a possible 16 cantrips. The restraining factor here is deck space. In a deck like Storm, you can play the full amount, since you can keep the interactive cards to a minimum. All of your deck is mana, kill cards, and ways to find enough of each. A fair deck like Delver can’t just ignore what the opponent is doing, you have to interact. This takes actual cardboard slots in your deck and usually, since you will have to interact early and often, quite a few of them. When your deck needs to contain mana, threats, and interactive cards, the space for cantrips is limited even though playing more cantrips will let you play fewer copies in each category, due to velocity. Also there is a threshold where the loss of tempo is too hard to overcome when the cantrip numbers get too high, since when your cantrips mostly just draw new cantrips you aren’t actually doing anything.
So is playing some number of cantrips correct no matter the deck? No. A good example of a deck that doesn’t include cantrips despite the ability to cast them is UWR Control. The deck has a lot of taplands and high casting-cost spells, so it usually falls behind on tempo anyway. Adding cantrips to such a deck would therefore be strategically unsound. The idea behind that deck isn’t to have more, less powerful spells. It mainly wants to trade evenly with the opponent and get ahead by having more powerful expensive cards as well as lands with spell-like qualities such as Celestial Colonnade or Tectonic Edge.
A deck like Splinter Twin combo is in an interesting position since it would greatly benefit from a higher velocity and doesn’t have a large portion of the deck locked in with threats. It does however play somewhat like a control deck a decent amount of the time. What that means to me is that it has high-casting-cost cards, a tendency to be favored in the late game, and lots of interactive cards. This is a natural fit for the deck, since the controlling game plan has synergy with the combo in that you want to keep mana open for the end-step Deceiver Exarch/Pestermite anyway, mana that can also be used for interactive cards should the need arise. The interactive counterspells can also be used to protect the combo once assembled. So Twin demands a delicate balance to achieve a healthy velocity and enough of each resource to be able to execute and protect the combo once assembled. For this reason something between 4 and 9 cantrips is ideal depending on the exact build.
In this section I would like to focus on the sequencing of cantrips. The cantrips in question are:
Gitaxian Probe: Mana-neutral (in a world of Bolts, fetches, and shocklands, I think the term “free” is used too liberally), helps plan the next couple of turns, triggers Ascension, Ascendancy, Swiftspear, or Pyromancer a turn early. Turns Snapcaster into Silvergill Adept.
Serum Visions: Sets up the next turn the best and digs deepest looking for specific cards.
Sleight of Hand: Best option for finding a card or type of card.
Gitaxian Probe has something of a special place, since it doesn’t require mana to cast. Therefore casting one before playing your first land to glean the information you need in order to perfectly sequence the rest of your plays can be quite good. That gets worse if you play against something like Burn, but can also save you life when you then play a tapped Steam Vents for example.
There are good reasons to hold onto Probe, though. Most notably if your hand develops naturally, meaning the information will not alter your plays in the first few turns, holding it to power up a turn-2 Young Pyromancer or a drawn Monastery Swiftspear can be good, as well as getting more information later on before you have a critical decision to make, say to find out if they have a counter for that Jeskai Ascendancy on turn three or not. Also remember that you can draw a card that you scryed to the top with a Serum Visions immediately. As will be the underlying theme throughout: when in doubt, hold it for another turn.
This is especially true for Serum Visions. I have found that it is quite often not correct to play Visions on turn 1 even if you don’t have another play. Say your hand has multiple fetchlands, a Pyromancer, and a Bolt or two. Since you are not looking for anything specific like a second land or a threat to play and already have an answer to his or her first play, I would hold the Visions for multiple reasons: First, saving the life on a tapped Steam Vents might be quite relevant depending on the matchup. Second, if your opponent has a Tarmogoyf in hand but no sorcery of his own, playing the Visions could put it out of Bolt range. Third, scrying and then having to use a fetchland next turn negates part of the scry (even if you only keep one card, the card you bottomed will get shuffled back in. If you keep both it is obviously even worse). And fourth, the more information you have the more informed your scry decision will be. Usually if you will use all of your mana for the foreseeable future and/or are looking for something specific right now play the Visions, otherwise hold it.
Sleight of Hand plays similarly to Serum Visions in that more information will make your decision easier. It is, however, different in that the effect on fetching afterward is less profound and you get immediate access to the card you take. In general it’s fine to fire it off when you can. When you have both Visions and Sleight in hand and don’t need a specific card this turn, cast Sleight first and Visions after if you can cast both, and Visions first if you can’t. If you do need a specific card this turn, reverse the order.
Thought Scour is the only one of these that is an instant, so this is a fine end-of-turn play when you need to hold up countermagic or removal. Since no decision-making is required in the resolution and it doesn’t change your plays, mostly you can just cast it whenever possible. The interaction with Visions is such that since you usually want to target yourself it is better to cast Scour before casting Visions, although if you are desperate for a certain card you can play Visions and then target the opponent to draw the card you scryed to the top.
Some niche uses are resolving a Delver trigger and then using Scour if you don’t want to draw the card (or if you really don’t want an opponent to draw the card they revealed off their Delver, target them before they draw). If you know something is on top that you a) don’t want to draw or b) actively want in your graveyard (examples are against Goblin Guide, or see a card like Past in Flames or Fatestitcher when you scry) you might also want to use Scour in your upkeep. If the opponent scryed something to the top and you win if they don’t draw a specific card right now, you can also use it to target them.
In general I see a lot of people casting their cantrips like they might a creature, just to get it out of their hand. This is mostly good practice for creatures, since they give you more value the longer they are in play. For cantrips involving decisions this is generally inverse, so you need to actually have a reason to cast it. Don’t cast your Serum Visions when you need both lands and all kinds of spells equally. Think of possible scenarios that might happen later on where you might rather still have it in hand than a random card. Maybe your opponent will cast a Thoughtseize and see your hand of Serum Visions, Gitaxian Probes, and lands. By taking one of your cantrips, he essentially trades tempo, life, and a card for what is essentially a random card. If you had cast those, you might have a juicy target for him to take now. By keeping them, you get a better idea of what you need and what you might want to bottom.
To sum it up, don’t play your cantrips because you can, play them to achieve something specific. Mana efficiency is an overrated concept. It should not be your reason to cast your spells, especially not your cantrips.
And now for some examples! Here are some starting hands, see how you would sequence the cantrips to sculpt your draws:
(In all instances this is a game-one situation where you are on the play vs. an unknown opponent.)
Playing: Splinter Twin
Play Tarn, pass (if they play a Swiftspear, Delver, Goblin Guide or some such, fetch for Mountain and Bolt it, otherwise get Vents tapped). Against a Hierarch it is likely worth the 2 life to get an untapped Steam Vents and Bolt it end of turn. Depending on your draw and their play, it might be correct to then go ahead and play the Visions after cracking the fetchland. A good reason would be looking for a third land or something like land+Twin if you drew a land so you can go turn 3 Exarch turn 4 Twin.
You might be tempted to play it turn one to look for something like a Remand, but there aren’t a lot of cards you would need to Remand on turn two and Remand might be actively bad in the matchup. Also, you have two lands and are going to need four and a Twin to win the game, so what are you going to bottom while looking for that Remand? Always think of what you need and what you wouldn’t. In this case neither is certain, so hold it.
The main goal with this hand is to find and play Ascension ASAP. What the opponent has will not greatly affect our game plan, so we do not need to Probe on turn one. It is tempting to do so since we have two, but since our game plan will likely revolve around Ascension (Electromancer/Past in Flames games involve multiple Rituals, of which we have none) we will want the second one to trigger that. So the play is fetch for Steam Vents, play Visions. Bottom basically anything that is not an Ascension.
Assuming we don’t find it, the second turn will be Probe, Manamorphose (R,U) and if we still haven’t found it play Thought Scour targeting yourself and pass the turn. You might think that the first Probe is free to play since we will need one in the graveyard anyway to trigger the Ascension, and you get information on what the opponent plays. This is not so, if we find an Ascension to play on turn two without playing either the Manamorphose or the Probe we can then lead on Thought Scour and maybe mill either of those cards to get a free counter on the enchantment. Also with multiple free spells in hand the Ascension might be a lot more threatening then it looks with just Serum Visions in our graveyard, so the opponent might play something like a ‘Goyf to put on pressure first, thinking you will need a turn to get it online before destroying it with an Abrupt Decay. Playing the Probe could actually give the opponent more relevant information than you get by looking at his or her hand.
This is a hand that doesn’t have one correct line of play, since it heavily revolves around the first couple of draw steps. Basically, you want to play the Pyromancer on two and then Probe to get the trigger and the information on what to hold Remand for, what you might need to scry to the top, and what you will not need. Then keep Remand up on turn three while ideally playing Serum Visions beforehand, and if they play nothing you want to Remand, still be able to Thought Scour at end of turn for another token.
This line of play necessitates a third land in the top three cards however. If we don’t draw the land while also needing to interact, we might be in a disadvantaged position. The line I would go for is to assume that you will draw the third land in time, since if we don’t it’s not too bad (we can still Remand on three then untap and Visions and Scour to find it while netting Elementals).
So the play is: turn one play Misty and pass. End of turn crack for tapped Steam Vents. Untap, play Tarn, fetch Island, play Pyromancer and Probe. Next turn, untap, play the third land (if fetch, fetch first) play Visions, and hold up both Remand and Scour. I wouldn’t fault anyone who cracks for Island off Misty on turn one here to cast Visions and look for the third land. This sacrifices value for a bit of security and is up to preference. The middle of the road play is to crack the Misty for Island at end of turn and cast Thought Scour. This increases the odds of hitting the land slightly while not wasting the value of the scry, just the 1/1 from Pyromancer.
This line is one of the reasons to play Misty over Tarn on turn one, since if you play against an aggressive deck you might want to fetch Island and Scour (since the Pyromancer has low chance of survival against a Lightning Bolt deck, you thereby increase your odds of finding the land while sacrificing less) and then need the Tarn to fetch Mountain to preserve your life total. The other reason is of course that Scalding Tarn is the best land in the deck and if you are sure you don’t need to fetch a Mountain, it is better to lead with Misty. Now, if you draw Island you can play that on turn two if you fetched Vents and still have the Tarn to possibly fetch a Mountain later on which Misty wouldn’t make possible.
Another possibility is to Probe on turn one for life to maybe draw a 1-drop, but since that line doesn’t guarantee value off the Pyromancer and there aren’t a lot of things you would need to Remand on turn two on the play, I would only recommend this line on the draw, since the information gained by the Probe would then tell you if you need to hold up Remand on two or if it’s safe to run out the Pyromancer.
I hope this has been helpful and that it helps you view Modern cantrips in a new light and maybe milk some extra value out of them in the future. Having delve spells in your deck might have been a strong incentive to play cantrip-heavy decks, but it wasn’t the only one and those types of decks still retain a lot of their punch, despite the bannings.