Part I: Intro
Posters have hitherto only selectively referenced replays and usage stats from RUWC 2022; the point, however, is to accurately present the data such that they can inform rather than misinform our decision-making. Such is the purpose of this essay.
As alluded to by the posts above, we decided to conduct an experiment during RUWC 2022. During this tour, we had two different SMRU slots - one regular SMRU slot and one suspect slot (SMRU + Talonflame). The idea behind this design is simple: we have two tiers that are equal in all respects but one - the presence/absence of Talonflame - hence any observable differences between the tiers can be attributed to the presence/absence of Talonflame. So, all we need to do is to gather data on SM vs SM + Talon, compare them, and then infer the effect of Talon. Then, in the final analysis, we may discuss whether these changes are for good or for bad.
Still, this leaves two important questions unanswered. First, what kind of data should we gather? Usage stats? Replay analysis? Thick description of what it is like to build and play with Talon? While each of these approaches can be fruitful, they also have distinct weaknesses. Firstly, the usage stats and replays are drawn from a small sample. Secondly, by naming the slot “SM RU + Talonflame”, we are implicitly suggesting that Talonflame is broken, thereby incentivizing people to a) use Talonflame more than what they would if the tier had a neutral name, and b) overprepare for Talonflame and thus inflate the usage of certain mons, which in turn would make Talonflame seem worse than it is. In other words, the usage stats and replays may not be representative of what a long-term Talonflame meta would look like. Alternatively, we can avoid these issues by simply asking people to describe what it was like to build and play with Talonflame, akin to what cyanize and LBN did above, but this introduces another source of bias: selective memory. We are, after all, asking people to describe what it was like to play a format that lasted for a few weeks, two years ago. In other words, while it can be useful to gather data from RUWC 2022 to study the effect of Talon, there is a significant risk of biased inferences, so the results should be interpreted with humility.
The second unanswered question is perhaps even more important: why do we want to free Talon? What effect do we want it to have? Suppose cyanize is correct that freeing Talonflame would make the dominant core of Donphan + Metagross (Registeel) + Golisopod worse and as a corollary improve hyper offense, sun, and sand; would this be a good thing? Put differently, do we want to see more of these offensive playstyles and less of classic balance and classic bulky offense? Conversely, suppose that freeing Talonflame only has a negligible effect on the tier; would this be sufficient to free it? Or does it need to have some form of impact on the tier to justify freeing it?
My tiering philosophy is as follows: I want to facilitate the creation of a format that is competitive and fun/interesting. To this end, I think a good tier should a) be centralized around a set of mons that have reliable checks and counters (to ensure that it is competitive), and b) leave room for creativity with respect to choice of play style, individual mons, sets, and items (to ensure that it is fun). Crucially, SMRU satisfies both a) and b). This has three important implications: 1) I am fine with keeping SMRU as is. 2) If Talonflame has a negligible effect on the meta, such that it remains both competitive and dynamic, I would not be opposed to freeing it. 3) Conversely, if the presence of Talonflame decentralizes the tier, for instance by improving sun, sand, and hyper offense at the expense of classic balance and bulky offense, I do not want to free it. Sure, this would increase variety even more, but it would also make it more difficult to build teams that are truly good – teams that have a decent match-up versus any given opponent – thereby making the tier less competitive.
So, how does Talon affect SMRU? To answer this question, I will analyze usage stats and replays from RUWC 2022 and describe my own experiences with playing and building in the Talon format.
Part II: Talonflame’s effect on usage
Below, I have grouped the usage stats according to typing, with SMRU vs SMRU + Talonflame usage side by side such that it is easier to compare them. Raw stats can be found here:
https://www.smogon.com/forums/threads/ru-world-cup-iv-usage-stats.3706718/
Flyings usage SMRU vs SMRU + Talon
| Flyings | SMRU | SMRU + Talon |
| Talonflame | 0% | 26.19% |
| Mandibuzz | 28.57% | 11.90% |
| Noivern | 26.19% | 26.19% |
| Sigilyph | 7.14% | 9.52% |
| Xatu | 11.90% | 2.38% |
| Mantine | 0% | 7.14% |
waters usage SMRU vs SMRU + Talon
| Waters | SMRU | SMRU + Talon |
| Blastoise | 23.81% | 21.43% |
| Golisopod | 21.43% | 4.76% |
| Slowbro | 9.52% | 21.43% |
| Slowking | 7.14% | 4.76% |
| Vaporeon | 7.14% | 4.76% |
| Milotic | 7.14% | 9.52% |
| Seismitoad | 4.76% | 2.38% |
rocks usage SMRU vs SMRU + Talon
| Rocks | SMRU | SMRU + Talon |
| Rhyperior | 9.52% | 30.95% |
| Diancie | 7.14% | 11.90% |
| Aerodactyl | 7.14% | 7.14% |
| Gigalith | 0% | 7.14% |
| Tyrantrum | 0% | 4.76% |
elecs usage SMRU vs SMRU + Talon
| Elecs | SMRU | SMRU + Talon |
| Raikou | 9.52% | 14.29% |
| Vikavolt | 0% | 2.38% |
| Ampharos | 0% | 0% |
steels usage SMRU vs SMRU + Talon
| Steels | SMRU | SMRU + Talon |
| Registeel | 28.57% | 16.67% |
| Metagross | 28.57% | 28.57% |
| Forretress | 19.05% | 21.43% |
| Bronzong | 16.67% | 14.29% |
| Escavalier | 4.76% | 7.14% |
grounds usage SMRU vs SMRU + Talon
| Grounds | SMRU | SMRU + Talon |
| Donphan | 21.43% | 19.05% |
| Flygon | 28.57% | 23.81% |
| Nidoqueen | 11.90% | 7.14% |
| Zygarde | 4.76% | 2.38% |
| Steelix | 2.38% | 0 |
| Torterra | 0 | 2.38% |
grasses usage SMRU vs SMRU + Talon
| Grasses | SMRU | SMRU + Talon |
| Sceptile | 19.05% | 11.90% |
| Roserade | 16.67% | 16.67% |
| Virizion | 11.90% | 23.81% |
| Shaymin | 9.52% | 9.52% |
| Dhelmise | 7.14% | 7.14% |
Let us begin by looking at cyanize’s claims that a) SMRU is dominated by don metapod (Donphan + Metagross / Registeel + Golisopod), b) don metapod’s dominance harms sun, sand, and hyper offense, and c) Talonflame punishes don metapod, thereby improving sun, sand, and hyper offense. If these claims are correct, we would expect that Donphan, Metagross, Registeel, and Golisopod would see a very high usage in the regular SM format, whereas they would see a significant decrease in usage in SM + Talonflame. Further, we would expect that sun and sand see more usage in SMRU + Talonflame than in the regular tier.
Interestingly, the usage stats don’t perfectly reflect this. As for a), cyanize is right that don metapod is dominant: each of its members saw more than 20% usage in the regular SM slot. Yet with respect to b), we find mixed results. On the one hand, Golisopod’s usage crashed from 21.43% to 4.76% and Registeel’s usage dropped from 28.57% to 16.67%. But on the other hand, Donphan’s and Metagross’ usage were unaffected by the presence of Talonflame. In a similar vein, claim c) is only partly supported by the data. Sure, sand did indeed see more usage in the talonflame format, with Gigalith being brought three times (7.14%), compared to never being brought at all to the normal format. However, sun saw less usage in the Talonflame slot with Ninetales’ usage falling from 9.52% in SMRU (brought four times) to 4.76% in SMRU + Talonflame (brought two times). In other words, Talonflame seems to have hurt sun rather than helping it, according to usage stats.
Leaving aside cyanize’s claims, let us proceed to look at winners and losers in a broader perspective. First of all, the main benefactors of Talonflame’s entry to the tier were Talonflame itself, Rhyperior, Slowbro, Virizion (!), and Raikou. Talonflame’s usage rose from 0% to 26.19%, Rhyperior’s usage increased from 9.52% to 30.95%, Slowbro’s usage rose from 9.52% to 21.42%, Virizion’s usage increased from 11.9% to 23.81%, and Raikou’s usage increased from 9.52% to 14.29%. Conversely, the mons that suffered the most from Talonflame’s entry were Golisopod, Mandibuzz. Registeel, Sceptile, and Xatu. Golisopod’s usage crashed from 21.43% to 4.76%, Mandibuzz’ usage stooped from 28.57% to 11.9%, Registeel’s usage dropped from 28.57% to 16.67%, Sceptile’s usage decreased from 19.05% to 11.9%, and Xatu’s usage fell from 11.9% to 2.38%.
The above are the most important changes, but we also saw other smaller changes in usage. Diancie, Gigalith, Mantine, and Tyrantrum saw a moderate increase in usage, whereas Flygon and Nidoqueen saw a modest decrease in usage. More specifically, Diancie’s usage rose from 7.14% to 11.9%, Gigalith's usage increased from 0% to 7.14%, Mantine’s usage rose from 0% to 7.14%, and Tyrantrum’s usage increased from 0% to 4.76%, whereas Flygon’s usage fell from 28.57% to 23.81% and Nidoqueen’s usage decreased from 11.9% to 7.14%. Finally, we saw some other even smaller differences in usage, but these changes are so small that they’re probably just coincidental (we’re talking about mons being brought one time more/less).
TL;DR: The usage stats indicate that a Talonflame format would be similar to the present meta. The major changes include significant increases in usage for Rhyperior, Slowbro, Raikou, and Virizion, as well as significant decreases in usage for Golisopod, Mandibuzz, Registeel, Sceptile, and Xatu. Interestingly, while both Diancie, Gigalith, and Tyrantrum saw a modest increase in usage, their usage did not surge. Further, Aerodactyl’s usage remained the same, whereas Ampharos was never brought to either meta. This indicates that, rather than having an especially warping effect on the meta by forcing people to use niche mons, the introduction of Talonflame simply gave people an incentive to use some already solid mons (Slowbro, Rhyperior, Raikou, Virizion, and Talonflame itself), at the expense of some other solid mons (Golisopod, Registeel, Mandibuzz, Sceptile, and Xatu).
Part III: Replay analysis
Talonflame was brought in eleven games during RUWC 2022. Unfortunately, three of the replays were lost. We do, however, have eight replays that are available, so let us go through each of them in turn.
Game 1: Saurav vs Crossbow
https://replay.pokemonshowdown.com/smogtours-gen7ou-644887
Preview: Talon has an awkward match up. It's being used on hyper offense, so it’s obvious that it’s going to be an offensive SD set, with either Fly Z, Steel Z, or Grass Z. Steel Z could dent Rhyperior for Lycan-D, while Grass Z could outright remove it, but usually Talonflame opts for the more reliable set with Fly Z.
In practice: Talon turns out to be SD Fly Z Steel Wing, which ends up being an outright liability. The only thing it does is to invite in Rhyp three times, which in turn gets to do 90% to Aero, set SR (vs a team with three SR weak mons), and kill Talon.
Saurav vs Feli
https://replay.pokemonshowdown.com/smogtours-gen7ou-644393
Preview; In this game it’s not as obvious what kind of set Talon will be running. SD Talon would be pretty bad here as well, given that it ran into another Rhyperior, but a fat pivot Talon looks okay since it can punish non-water moves from Pod, Spin/Gyro from Forre, threaten out Rose, and burn Rhyp with Wisp.
In practice: Talon switches in two times (each time after a sack). The first time it threatens out Golisopod and burns Rhyp. The second time it dodges a Fire Blast from Houndoom and does 75% in return. The opponent then switches to Rhyp on the following turn, tanks the incoming BB, and kills Talon.
Crossbow vs Feli
https://replay.pokemonshowdown.com/smogtours-gen7uu-642456
Preview: Talon couild be any given set on this build. SD would be terrible, given that it ran into a Gigalith, but a tanky set could punish U-turn from Flygon and any given move from Escav. It could also potentially 1v1 Noivern and Sigilyph but it depends.
In practice: Taunt Roost Talon is used to switch into Noivern, after Noivern wasted the Z-move into Slowbro. It is also used to catch Escav, before then doubling to Passimian to make progress. Beyond this, it does nothing.
Clem vs GXE
https://replay.pokemonshowdown.com/smogtours-gen7uu-644475
Preview: Difficult to tell what set Talon would be. SD Fly Z has potential to kill everything if piloted right, but if it’s not Z, or if has already used it, then Vaporeon will sit on it with ease.
In practice: Offensive lefties U-turn Talon is used to get Raikou in versus Vaporeon, before then Calm Minding and removing the opposing rocker (Metagross) with a Z-move. Later in the game, Talon switches into Shaymin a couple of times and U-turns into Blastoise. In the end, however, the team ends up losing because all remaining pieces (after Virizion and Raikou go down) are walled by Vaporeon.
Btboy vs DuGuo
https://replay.pokemonshowdown.com/smogtours-gen7uu-644288
Preview: Rhyperior + Heal Bell is probably the worst possible match up for Talonflame. A rogue Grassium Z could maybe snipe Rhyp, but it would have to be played well. Fat Talon could burn Rhyp and then hard switch or U-turn into Tang/Pass/Mandi but in all cases the opponent simply switchess to Florges and clicks Heal Bell. Further, if Rhyp is Toxic, then Mandi won’t even be able to keep SR off versus burned Rhyp.
In practice: Talon switches into Salazzle, takes half and a poison, and has to Roost on the following turn as Rhyp switches in. Talon then dies to Stone Edge after taunting Rhyp.
DuGuo vs LBN
https://replay.pokemonshowdown.com/smogtours-gen7uu-644399
Preview: Not sure what the Talon set is. SD could potentially get in versus Tort, Rose, or Forre and make some progress, but it needs watch out for Stone Edge Tort in addition to being unable to 1v1 Aero, Milo and Necrozma. Wisp + Taunt would be bad since Milo just sits on it. A fat U-turn set looks decent, since it can switch into Tort and and Forre and pivot into Virizion versus incoming Milo or Rhyp versus Aero, but it won’t be amazing.
In practice: Talon comes in four times. First, it switches into Forre, gets 40% off on Milo, and U-turns out on incoming Aero predicting the Z-move. Next, after sacking Cress, Talon has to come in versus Rose and click Roost to stay healthy. Third, after sacking Blastoise to Rose, Talon switches in and does 35% to Aero, before having to sack Virizion. Finally, Talon switches into Torterra and trades most of its health for koing Tort. The opponent then goes into Necrozma and cleans up the game.
eifo vs LBN
https://replay.pokemonshowdown.com/smogtours-gen7uu-643055
Preview: Difficult to tell on preview what set Talon will be. LBN has to watch out for SD Steel Wing Fly Z, which has the potential to clean his team. Fat Talon looks okay, since it can sort of handle Pangoro (or at least punish it with Flame Body), punish Flygon’s U-turn with Flame Body, and forces Bronzong to switch into Diancie, which it can exploit by U-turning into Metagross and clicking Mash into incoming Blastoise.
In practice: Talon switches into CB Drain from Panda with SR up and dies. However, it also gets off a Flame Body burn which ends up winning me the game. I would have been disfavored in the endgame without the burn.
eifo vs yay
https://replay.pokemonshowdown.com/smogtours-gen7uu-642736
Preview: Classic don metapod feat Talon instead of Scarf Houndoom. This style relies on momentum, so we can rule out Taunt Roost BB Wisp, but it could be anything else. At first glance, SD Steel Wing looks great since it kos anything at +2. However, every mon on the opposite side could potentially OHKO it, so it will have to be careful. Likewise, a fat U-turn set would be solid, since it can punish Mash from Meta and U-turn from Flygon, but it needs to watch out for Elec Z/CB Tpunch and Stone Edge. Still, it should make yay think twice before spamming Mash/U-turn. Moreover, burning Diancie, or even luring it in to take hazards and U-turning out into Meta/Rose will be useful, given that this team hates SR and that weakening Dia will go a long way towards keeping them off.
In practice: Talon never comes out. I win the game without using it.
TL;DR: Talonflame looks better on preview than in practice. It was always possible to think of some hypothetical Talon set that could break anything at preview, but in practice we did not see a single game in which Talonflame had a dominant appearance. Set-up sets were particularly underwhelming, with both SD and Bulk Up failing to do anything in any of the games where they were brought after running into Rhyperior. In fact, based on the replays, the best set seems to be fat Flame Body Talon, since this set not only provides defensive value by punishing Knock, Turn, Mash, and functioning as a decent grass/fighting resist, but also has the ability to burn Rhyp/Aero upon switchin and U-turn out to retain momentum.
Part IV: My experiences with building and playing with Talonflame during RUWC
To begin with, Talonflame made building more difficult for me. It had several different sets that it could use, some of which had different checks, so how could I be expected to build teams that reliably answer it? For instance, what am I to do if my team relies on Slowbro/Milotic + Diancie + Scarf (Stone Edge) Flygon to answer Talonflame and I run into a Flyinium Z Steel Wing Talonflame? Or what if my Rhyperior/Aerodactyl/Tyrantrum runs into Talonflame and it gets burnt by Wisp? Worse yet, what if my Rhyperior gets sniped by a Z Solarbeam? Do I have to use Rhyperior + Heal Bell or Rest Talk Ampharos-M on every team? No, thankfully not.
As showcased by the replays, Talonflame turned out to be more manageable than I had feared, for two reasons: Firstly, the SR weakness forces the Talon user to keep SR off the field, which means that they must waste turns on removing hazards while the opponent can use those turns to make progress. Secondly, although Talon can theoretically run 10 different moves and use 4 different items, it will usually either be an offensive Gale Wings set with SD+BB+Blitz+U-turn @ Flyinium Z or a bulky Flame Body set with BB+U-turn+Roost+Taunt/Wisp/Toxic @ Helmet/Lefties/Fly Z, as these two sets perform the most consistently. Sure, it can sometimes be worth it to use lures such as SD with Steelium Z or Grassium Z, but this comes at a significant opportunity cost in that you forego the defensive utility of a Roost Talon, the ability to break fat waters such as Slowbro and Milotic with Fly Z, and the option of pivoting out with U-turn. In other words, if you do not manage to lure the Diancie/Tyrantrum/Aerodactyl with Steelium or the Rhyperior with Grassium, your Talonflame will perform rather poorly, as is evidenced by the replay in which the SD Steel Wing Talonflame flails into Rhyperior.
Still, this is not to say that Talonflame had no effect on teambuilding. Firstly, SD Talon was annoying to deal with for teams that relied on Meta/Regi + Bro/King/Milo as the switchins to flying moves, which made me opt for rock types more than I otherwise would, especially Diancie, Rhyperior, and Aerodactyl. Secondly, Flame Body Talon punishes Mash/BP/Ice Punch from Meta, Knock from Don, Seismic Toss from Registeel, First Impression/Leech Life/Knock from Golisopod, dual stab synthesis SD/CM Virizion, Drain/Knock from Toxicroak, and U-turn from pivots such as Flygon and Passimian. As a result of this, I always used a water move on Golisopod, a rock moves on Flygon, Donphan, Virizion, and Passimian, opted for Electrium Z and Explosion more often than usual on Metagross, and became slightly more reluctant to use Registeel. Apart from this, my building stayed the same. I continued to build nothing but standard balance and bulky offense, albeit with some slight tweaks. And I did so to great success, winning all but one game.
TL;DR: Talonflame punishes Knock, U-turn, Mash spam, and builds that rely on steels as flying resists, which led me to use Diancie and Rhyperior more at the expense of Registeel, while also opting for rock moves more often on Flygon, Donphan, Virizion, and Passimian, in addition to using Electrium Z and Explosion more often on Metagross.
Part V: conclusion
Whether we ought to free Talon depends on two central questions: 1) What effect do we want Talonflame to have? 2) What effect do we expect Talonflame to have? Personally, I already love SMRU, so I do not want it to change too much. A good tier, in my opinion, is a tier which is centralized around a set of mons that have reliable checks and counters, while also leaving room for creativity in the builder. SMRU is characterized by both of the above, so I do not share cyanize’s worry about the dominance of don metapod. Consequently, I would not support freeing Talonflame if, as cyanize claims, Talon would make sun, sand, and hyper offense better at the expense of classic bulky offense and balance. However, this does not seem to be the case. During RUWC 2022, sand did admittedly see a boon, but sun and hyper offense did not get any better. Most importantly, classic bulky offense and balance remained dominant, as is evidenced by Rhyperior and Slowbro’s increased usage. In the end, the main effect of Talon was to incentivize the usage of certain solid mons (Rhyperior, Slowbro, Raikou, Diancie, Tyrantrum, Aerodactyl, and Gigalith) at the expense of other solid mons (Registeel, Golisopod, Mandibuzz, Sceptile, and Xatu), and to opt for slightly different coverage moves on some other solid mons (electric/rock coverage on Meta, Don, Flygon, Pass, Viriz). Ergo, I would not be opposed to freeing Talon, but I would also be fine with keeping the format unchanged.
Addendum
Before I go, I must end this with a caveat. This analysis is founded on an inferential logic whereby I, based on the observable differences between one format (SMRU) and another format (SMRU + Talonflame), infer the effect of Talonflame. To this end, I gathered data in the form of usage stats, replay analysis, and a subjective description of what it was like to play the Talon format, before then drawing inferences based on these data. However, as you may recall, these data are flawed. Firstly, the usage stats and replays are gathered from a small sample size and from a format which had a distorting name (SMRU + Talonflame), which means that they may not be representative of what a long-term Talonflame meta would look like. Secondly, while relying on data in the form of my own subjective experiences with playing the Talon format avoids this issue, it introduces a new source of bias: selective memory. In short, there's a significant risk that the data from which my inferences are drawn are biased, and, as a result, that my inferences are biased. In other words: I may be wrong. My conclusions may be faulty. In the end, then, we are left with uncertainty, but hopefully this analysis can spur some more thought and discussion.
Thank you to my friend
cyanize for inspiring this post and for reminding me to add the caveat at the end.