It's hard being the Modern data guy. There's a ton of data entry every week and there really aren't shortcuts available. At least there aren't when I want to actually track how decks evolve. Which is essential to actually telling the story of the metagame. Decks diverge, merge, and reinvent themselves over time. If I didn't watch it all happen every week, I'd miss the actual metagame shifts I'm tracking.
The Analyst's Lament
As the desired granularity of data increases linearly, the difficulty of automation increases exponentially. It is relatively easy to get a scraper to collect data from various sources and collect it into a single document. Programming the scraper (or a separate program) to then automatically sort and categorize the data is very hard. If you're looking for something very specific, it's not difficult to tell the program "Exactly This or Nothing." You do have to know exactly what that thing is to begin with and not care about anything else, though.
In all other cases, having collected the data, you have to program in how the data is to be sorted and categorized, and this is where the problems begin. You have to set limits on the acceptable variation within the data. If the variation is known, that's no problem. When it's unknown, you need to set it yourself. Set it too low and exclude data you don't want. Set it too high and there's no point in sorting it in the first place. It gets to the point where the effort needed to automate the process is greater than doing it manually.
As Pertains to Magic
Which is the problem I have every month. When I first started doing this, I did try to use bots and scrapers to ease the workload. It backfired. I always ended up with so many decks marked "Other" that needed to be sorted into their correct archetypes that I wasn't really saving myself any work. I didn't have to do all the data entry, but I did have to spend so much time auditing the bots work that it wasn't worthwhile.
Consider the problem with contemporary decks. UR Murktide has a very recognizable deck design. Any reasonably informed player can look at any decklist in that MTGGoldfish link and agree that the decks are undeniably UR Murktide. But how would I program a bot to do that? The decklists have a fairly standardized list of cards in them, but there are plenty of flex slots. And the land base is all over the place.
If I set the heuristic at 10% variation, that's a 6 card difference from some standardized list. That will get many of the right decks, but decks that shave on numbers to add more cards will get left out. So will the player who cheaped-out on the lands. Setting higher risks getting the wrong deck, like this UR Prowess/Blue Moon hybrid instead, which is 10 cards off from standard Murktide. I'm going to end up looking at almost every deck anyway, I might as well embrace it.
Because I'm looking at all the decklists as I enter them, I see how decks change and evolve. And sometimes that throws me for a loop and highlights how the metagame is changing. I'm seeing that happen right now in a particular archetype that is already a problem to classify. And its getting worse carries some worrisome implications.
Case in Point
There are 3 different decks that can be classified as 4-Color Omnath. They all play the MH2 pitch-elementals, are 4-Color decks, and are held together by Omnath, Locus of Creation. However, simply lumping them together is incorrect. There is a tribal version defined by Flamekin Harbinger and Risen Reef that is clearly unique. Distinguishing the Control vs Blink versions proves more difficult. The only consistent difference is the latter has Ephemerate, so that's the bright line. And it's worked perfectly.
A New Wrinkle
Until now, that is. Here are three recent Omnath decks. They are indicative of what I'm seeing in the data. What is the real difference between them?
There are continuities here. Every deck runs Omanth, Fury, Solitude, Endurance, and Teferi, Time Raveler. The main difference between Control and Blink continues to be Ephemerate. However, the distinction between Tribal Elementals and Blink is now also getting blurry, and that's something to watch.
A Sudden Shift...
I only noticed this change in the past couple weeks. Risen Reef has been in Blink lists before, but never consistently, nor was it as strategically integrated as it is now. The reference list features 18 elementals to trigger Reef, which is pushing it toward being a tribal deck. In fact, I may have classified it as Tribal Elementals in previous months and only noticed it being new in May.
The decision to embrace Reef does make sense. On its own, it fills a niche very similar to Ice-Fang Coatl, which is a staple in Control and had been for Blink. As 1/1 cantrip creatures, both are mainly there to bridge the midrange gap to the real spells and chump block. Both can be actual cards with some support, but Reef synergizes with the cards already played in the deck, while Coatl further strains the manabase. In exchange for costing more, Reef has the potential to get completely out of hand, while Coatl will just trade up at best.
...That's Inexplicable, But Scary
While older-style Blink does still make the data, its numbers are paltry compared to the Reef version so far. I don't know why this is happening, but it does appear that this Tribal Blink version is outcompeting the other Omnath variants across the board. I can't explain this development with certainty. I didn't notice this shift until May, but this configuration first put up results in April, so a shift in the metagame is not the cause. It might be that players just now remembered that such a move was possible.
More troublingly, it may be that there's no longer reason to distinguish between Omnath versions. The Tribal variant was primarily synergy- and tempo-based, while Blink and Control are about card advantage and power. Why not have it all? Is there an actual need to bother with control cards like Expressive Iteration or Counterspell when you're already playing Fury and Solitude? Better, use Reef to make up the card advantage from pitching the elementals. Then use Ephemerate to just roll in the value. Players already complain that Omnath is too hard for other fair decks to compete against. How can normal control or midrange decks hope to keep up with the flood of value this new variant generates?
More Literal Evolution
And then there's the fact that the Omnath decks are branching out. As far as I remember, they've always been midrangey value decks, but since Streets of New Capenna, they're moving into the combo space. And it's all thanks to Vivien on the Hunt
As I've mentioned before, Vivien and a three-drop translates into a combo kill. Some Omnath players have decided to embrace this fact and integrated Vivien into their deck. It does have the advantage of being in an 80-card deck, so the odds of drawing the combo pieces that need to be fetched are lower. However, again, I have to ask: is this actually a new deck? The noncreature spells look amazingly similar to the mix from the above control list.
And that's even more concerning than the move towards Tribal Blink. If 4-Color Omnath really can just do everything, is there any point in other decks trying to compete? To be clear, 4-Color Vivien is not a true combo deck like Tameshi Bloom. This is a Splinter Twin midrange/combo hybrid. I also haven't seen much evidence to say that the combo is actually good. Yet. However, the threat is definitely there.
There is nothing inherently wrong with 4- or even 5-color decks in Modern. If the decks required hoops to jump through or were doing something unique like Niv-Mizzet Reborn, there'd be no issue. However, once they start being piles of everything and anything good, they start pushing out alternatives and homogenizing. The evolutions in 4-Color Omnath I'm watching have the potential to do exactly that. I'm hoping that this is just the latest fad and isn't actually good. Because if they are actually good, I can't see it ending well.