Mel's W Rework: Riot Deleted a 'Signature' Mechanic and Forgot to Ship the Numbers

Two in the morning. Patch 26.22 open on one screen, cold coffee on the other. Three lines in and the coffee stopped mattering: Mel's W no longer reflects enemy projectiles. It destroys them now, and in exchange it grants a shield and movement speed to herself and nearby allies.

This is the mechanic Riot itself called Mel's "signature characteristic" in the same document. And two sentences later there is the small line most people scrolled past: Riot tried to preserve the reflect, and concluded it could not survive.

That sentence is the receipt. Everything else is commentary.

My stake, declared before it can be found: I am not a Mel main. My account has eleven Mel games on it and I dropped her after six of the losses. I have no personal grievance against the reflect. And one old error on record — in 2026 I wrote that Riot does not remove a champion's identity mechanic without first publishing the data behind the decision. 26.22 breaks that claim. There is no data here and the mechanic is gone. I am conceding it once, in this paragraph, and not returning to it.

I found the footnote before I found the headline, and the footnote was not written on announcement day. The notes refer to "continuous adjustment rounds" — meaning earlier hotfix cycles tried and failed to stabilise her. This rework is not a decision, it is the sum of three or four failed ones.

Context, because Mel arguments tend to drown in feeling. She is a mid-lane mage whose draft value rested on a single button. Reflect meant any projectile aimed at her could come back: burst, crowd control, a critical strike. Riot named two examples in its own notes — Seraphine's ultimate reflecting into a charm, Jhin's fourth shot reflecting a crit back into his own face. Both examples say the same thing: the opponent experienced dying to their own damage, and that experience became anger rather than respect.

But 26.22 is not only the W. The package carries four more changes and they need reading separately. Passive loses early damage and duration, gains health and armour scaling. Q shifts damage out of the first explosion into subsequent hits, with a slower projectile. E roots shorter at close range and longer at maximum range. R loses damage.

This is not a balance patch. It is the nationalisation of an identity. Riot removed the mechanic it had described as the champion's defining trait and paid for it in utility. On raw damage Mel goes down — passive, Q and R all cut. On utility she goes up — shield, movement-speed aura, health and armour scaling. The consequence is a role shift from burst counter-pick to sustain-enable mid. That is an identity change wearing the costume of a number tweak.

Look closely at the E change. Shortening the close-range root while lengthening the max-range root is not a punishment, it is a wager. It is skill-indexing: reward precise spacing, tax the panicked close-range engage. The player who used Mel as walk-up-and-lock-down has lost a tool. The player who lands max-range roots has gained one. A single patch, one champion, and two populations moving in opposite directions.

The passive change is quieter and more dangerous. Less early damage, more health and armour scaling — that does not adjust a number, it relocates a build path. Riot did not change Mel's damage profile so much as point her at different items. The day she shows up in a tankier build, nobody should be surprised; the direction was printed in the patch notes. And this is exactly where reworks slip, because multi-ability reworks almost never land correctly on the first pass. 26.22 is a first draft, not a settlement.

Now the part the whole column rests on: there is not a single number in this patch. No win rate. No pick rate. No ban rate. Riot's stated reason is frustration reduction — a feeling. Feelings can be legitimate design reasons. But if your entire justification is that a champion is unpleasant to face, ship the number that measures the unpleasantness. What was the ban rate? What share of games pre-emptively banned her on sight? What changed across the preceding hotfix cycles? None of it appears.

My core read is that this is design governance, not balance governance. The publisher is simultaneously rule-maker and stakeholder. When it decides which mechanic is anti-fun, that decision happens in a dashboard, not in a public filing and not in front of a committee. This is not new — Riot's pattern runs from the old Poppy and Yorick reworks to today: any mechanic that removes the opponent's agency will eventually be removed at the cost of the owner's attachment.

Which is where the weakest sentence in the document appears: the claim that these changes preserve Mel's identity. Calling something a signature characteristic and then deleting it, while claiming the identity survived, is a press line, not an analysis. The identity changed and the direction of travel is legible. Admitting it would have cost Riot nothing. Denying it costs credibility, because the player who spent three seasons on that one pick can read.

The winners and losers list is short. Projectile-reliant champions gain: Seraphine, Jhin, and by extension the Ezreal and Caitlyn poke-crit class. Teams that used to redraft around the threat of Mel get air. The losers are the players who held Mel as a stored counter-pick — her draft value drops, it does not vanish — and, most of all, the Mel one-trick with no second pick. At champion level this is the same event as an ageing cliff in a playing career: the day your greatest strength becomes your greatest liability.

One absence matters more than any of the above. There is no human name anywhere in this patch. No lead designer, no balance lead, no official who can be asked who signed. No pro player or team either, meaning the change carries no competitive reference point at all. Where a document has no human in it, there is no accountability in it. That is not a complaint, it is method — I look for names because structural analysis without a name is half-finished. Here the decision-maker is a structure: the publisher's dashboard. The alternative it declined was available and cheaper — weaken the reflect, or restrict it to a class of projectiles. Riot skipped that road and removed the button.

On timing, I will keep this as inference. The notes place 26.22 roughly three weeks after announcement. If that window lands inside or immediately before a major competitive calendar, a rework-level champion change opens a gap between the practice patch and the tournament patch. My experience says reworks reach the pro server a cycle late, so elite impact lags live impact. But I will say plainly: this article names no tournament, no team and no region. I do not speculate where all I hold is a date.

An editorial note that is really a habit. The source I read was the Vietnamese localisation — LMHT, with the ability rendered as Phản Quang Chiết Kính. This is not a regional story; it is a translation of global patch notes. Sitting in India, I see this pattern constantly: patch notes arrive here as verdicts, not as arguments. The number you could argue with never ships. Which is why my real question about 26.22 is not the W. It is where the data went.

Now the section where I argue against myself, because someone will do it anyway and would do it worse.

The strongest case against me is that Riot has the data and simply does not publish it — no obligation to print win rates in patch notes — and that removing an anti-fun mechanic is a defensible design decision on its own terms. If a champion's ban rate is 40 percent and that 40 percent exists because players refuse to tolerate her, the designer does not need your permission to act. That argument is strong and I accept it fully. My objection is not to the removal; it is to the procedure. The day you announce you tried to preserve an identity is the day you owe an accounting of what that identity cost.

And here is the sentence that would prove me wrong, written now rather than later: if within two patches Mel's ban rate halves from its prior level while her win rate holds between 48 and 50 percent, my impatience framing is wrong. It would mean Riot measured the frustration, acted on the measurement, and simply declined to show its working. In that case I will not delete this column. I will post the correction in the same thread, as I do every time.

Three testable predictions. First, Mel's pick rate spikes in the first two weeks after 26.22 goes live — that is the rework honeymoon and it happens every time. Second, the spike decays by week three or four, because once the burst counter-pick is gone a large slice of that population does not come back. Third and most important: if her win rate slips below 47 percent, Riot returns something within a patch — probably not the reflect, but damage in Q or the passive. I am predicting because I do not have the data, and when you do not have the data, a prediction is the only honest bet on the table.

The day Mel turns up in a tankier build, remember that no player invented it. It was written into two lines of the passive. Patch notes are where the future gets printed early — it is just that nobody reads them.