Murder Row

The whole run turns on ten seconds. Read off the combat log:

  17:48   Corrupted Warlock finishes a 1.5s cast, Curse of Doom lands on you
  17:48   ...and sits. It deals no damage at all while it is up.
  17:58   it expires and detonates: 629,489 to you, 643,590 to Ele Shaman
  17:58   you both die inside 0.4 seconds of each other
  18:11   the last of the other three dies

Two things make this the run rather than a bad pull. The bomb hit *two* people, so it splashes whoever is standing near the carrier — spreading is an answer even when nobody dispels. And it is not on the interrupt list: no one in the archive has ever kicked it, so dispelling or spreading were the only outs.

The pull was also the hardest thing in the key: the party's two heaviest eight-second windows, 15,482,759 and 12,385,876, land at 17:52 and 17:44 — immediately before the detonation.

What you did in those ten seconds was not nothing: Aura Mastery 0.3s after it landed, then Light of Dawn, three Flash of Lights, Beacon of Virtue, Holy Shock. You answered the raid damage. What never went out was a personal defensive — Divine Protection, which you pressed thirteen times elsewhere in this key. Against 629,489 on a 753,120 health pool (read off the log's advanced block), with Demonic Frenzy already ticking, that one press is the difference between a detonation and a death.

The grade docks you 3.2 points for `over-exposed: Murder in a Row x31 (peers 0)` — damage the tool believes comparable players avoided and you did not. Two independent checks say otherwise.

First, the name is not one spell. Five distinct ids in this log are called "Murder in a Row". `474545` is a marker that lands on all five of you for one to five seconds and deals **zero** damage to anybody. `474740` is the damaging bleed. Counted separately, the bleed landed on you twice and on nobody else — so "peers 0" is true, but it is true because the boss applied it to one person.

Second, position. The log carries x/y coordinates, which Arbiter never reads. At both applications:

  application     you      tank     nearest other
  17:17:48      27.0 yd   9.5 yd      5.5 yd
  17:18:30      24.2 yd  20.4 yd     15.1 yd

You were the furthest player from Zaen Bladesorrow the first time and second furthest the second. The boss reached out two dozen yards and picked the healer. (The boss's own position fix is ~5.8s stale — bosses only report position when they act — so treat the yardage as soft; the ranking is not.)

That is target selection, not standing in melee. Your `mechanics` should read near 100.

I first wrote that this axis was genuinely low. That was wrong, and it was wrong because I checked the easy objection (were you dead for the worst of it — no) and not the real one: was this much healing possible at all.

Measured across every full healer-key in the archive:

  this key demanded    236,031 /s of outside healing
  median key demands   105,450 /s
  next-hardest ever    151,204 /s   (Murder Row +13, 2026-08-29)

This key asked for 1.56x more than the hardest key on record and 2.24x the median. Against that you delivered 123,678/s of effective outside healing — the **highest of all 52 keys**, and 1.59x your own median of 77,716/s.

The ceiling test, done without using this key so it cannot be circular: take the best output rate you have produced in any OTHER key (120,359/s, the Altar of Fangs +13 this morning) and sustain it across the whole of this one. It covers 51%. You covered 52%.

To score at the corpus median coverage of 77% here you would have needed 180,882/s — 1.46x the highest rate anyone here has sustained.

So `throughput 42` is not a measurement of your healing. `healing_need` is an honest denominator — it already strips self-healing, other players' off-healing and overkill, and 66% of the party's 297,222,432 intake still survived those filters. The band on top of it is cut from a corpus whose demand is half this key. Nor is the wipe the explanation: those two pulls are 12% of the key's intake. The demand was relentless the whole way.

`response 45` is a different answer and partly real. Its window demand, 750,800/s, is also the highest of the 52 — so 31% reads worse than it was. But your absolute put-back rate in those windows, 230,150/s, ranks 15th of 52 rather than 1st, and the best rate ever recorded would have reached about 43% here. Call it twelve points of real room, not the fifty-five the score implies.

Which is its own finding: your sustained output was the best you have ever produced and your burst response was ordinary. On a key this relentless those are probably the same fact.

The party stopped 12 of 21 stoppable casts — 57%, the weakest of the four keys this week.

  Felfire Burst      5 of 5 leaked   (Unleashed Imp)
  Scathing Review    2 of 7 leaked
  Fel Rage           2 of 6 leaked

BM Hunter took zero interrupts against two expected and never pressed Tranquilizing Shot, which is most of the D- on that sheet.

In fairness to the kicks that were spent: at 17:21:32.812 Ret Paladin interrupted Fel Rage — 0.12s before Curse of Doom completed, from a different mob. There were more casts than kickers, and the one that mattered was not kickable anyway.