What it is. Every new unit, card, or piece of gear is strictly better than what shipped before it. The newest thing is always the best thing, so yesterday’s content is dead the moment today’s drops.

Why it hurts. Your library shrinks as it grows. Players don’t gain options — they gain a new mandatory answer and lose everything it replaced. The catalog you spent years building collapses into a thin, always-current meta, and returning players find their old favorites benched.

The smell

The new addition beats the old one on every axis — more damage, more range, more utility, cheaper cost. No trade sits on the table. “Better” is the only relationship two items ever have.

How it happens

• Numbers are the only lever. New content ships by nudging stats up, because +10% is faster to design than a genuinely new role. See antipattern-stat-inflation — power creep is its progression-over-time face.
• Sales pressure. The new set must feel worth buying, so it must feel stronger. Marketing demands a bump; the bump becomes a ratchet.
• No sunset plan. Content only accumulates. Nothing rotates out, so old and new must share one power budget forever — and the budget only tilts forward.
• Fear of “sidegrade” reads as boring. A designer worried players won’t chase a horizontal option reaches for a vertical one instead, killing pattern-meaningful-choice.
• The chase reward is the power. When system-reward-schedules pay out in raw strength, every drop must out-drop the last to feel like a reward.

The tell (check your own design)

• Can you name a living reason to pick a two-year-old item over this season’s? If the only answer is “nostalgia” or “budget deck,” you’ve crept.
• Line up every item that fills a role by ship date. Is the strongest one always the newest? That monotonic staircase is the fingerprint.
• Does a returning player’s old loadout still function, or is it auto-obsolete?
• Are your patch notes 90% “increased X” and 0% “here is a new thing X is good at”? Buffs-only is creep in slow motion.
• Would removing the newest tier collapse the meta to the previous newest? If power is a single tower, not a web of counters, system-counter-systems never got built.

The fix

• Sidegrade, don’t upgrade. New content should open a new axis, not raise an old one. Give it a strength that costs a weakness — trade damage for range, reliability for ceiling. This is pattern-risk-reward applied to the catalog itself, and it’s how system-build-diversity survives past launch.
• Rotate, don’t stack. Cap what’s legal at once. Seasons that cycle a pool keep the total power flat while the current mix stays fresh — the core discipline of system-live-ops-and-seasons. A rotating format is a sunset plan you build once.
• Design counters, not ceilings. Route new power through system-counter-systems so the answer to a strong thing is another thing, not a stronger version of it. A meta that self-corrects doesn’t need you to out-buff it.
• Balance the whole roster, not just the new tier. Buff the forgotten to meet the new instead of always raising the ceiling — system-unit-rosters and system-faction-asymmetry stay legible when the floor rises to the ceiling, not the reverse.
• Give power a shape, not a size. Tie progression to specialization — system-skill-trees, system-build-diversity — so “more powerful” reads as “more particular,” and system-progression deepens choice instead of flattening it.
• Sink the old currency of power. system-prestige-and-newgame-plus and reset-friendly system-meta-progression let you re-flatten periodically instead of inflating forever.

Twist seams

• A live-ops card game but power is a fixed budget every season redistributes (twist: conservation vector). New cards don’t add power; they move it — every buff is funded by a nerf, so the catalog stays flat while the meta churns. system-live-ops-and-seasons · genre-deckbuilder.
• An expansion roster but every new unit hard-counters an old staple and loses to another (twist: rock-paper-scissors vector). Strength is relational, not absolute — the newest unit is situationally best, never globally. system-counter-systems · genre-rts · system-faction-asymmetry.
• A gear ladder but each new tier trades a stat instead of raising it (twist: sidegrade vector). Late gear is different, not stronger — new builds open, old builds keep working. pattern-risk-reward · system-build-diversity.

Seen in / avoided in

Adjacent failure modes

• Creep that voids old content is a cousin of antipattern-solved-metagame — when only the newest tier is correct, the game solves to one deck fast.
• Numbers-only escalation with no new decisions is antipattern-stat-inflation; if it also floods the player with must-buy tiers, watch for antipattern-pay-to-skip and antipattern-grind-wall.
• A “new axis” that’s actually just a reskin is antipattern-false-depth — a sidegrade must genuinely change how you play, not the label on the stat.

Verify / guard

This is a design-time smell — catch it before handoff by auditing the roster for a monotonic power staircase (strongest-always-newest) and confirming each new item carries a real trade. See process-refine-and-handoff and the verification bar in docs/VERIFICATION.md. Balance proof and roster health live in the JUDGE half, not here — system-live-ops-and-seasons carries the rotation discipline that keeps this from recurring across seasons.