FAIRNESS — Mempool Invaders

NOTHING ON THIS PAGE IS A SUMMARY

Every number that decides a run is printed here in full. If a number in the game disagrees with a number on this page, the page is the bug report.

What this game is not. There is no paytable, no return-to-player figure, no jackpot, no meter, no near-miss, no loss-disguised-as-win, and no random payout of any kind. Not one value below is drawn at random. Every reward in Mempool Invaders is a pure function of what you did — which is why all of it can be published without weakening it.

1 · THE SAUCER TABLE

The saucer's value is never random. It is a frozen 15-entry table, indexed by a counter that advances on every shot you fire — hit or miss — and resets at the start of each wave. Entry direction is the low bit of the same counter: even enters from the right, odd from the left.

INDEX	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14
POINTS	100	50	50	100	150	100	100	50	300	100	100	100	50	150	100

Composition: five 50s, eight 100s, two 150s and exactly one 300, at index 8.

THE 23RD-SHOT RULE

Fire 22 shots and the pointer sits at index 8. Shot 23 lands on the 300, and every 15th shot after that does the same — shots 38, 53, 68 and so on. The rhythm a wave gives you is 22 · then 14 · then 14. The counter is drawn live in the HUD as a 15-tick ring with the 300 marked, because a mastery layer that requires a 1980 strategy guide is not a mastery layer.

Every SHOT transaction records the table index it advanced to, so this whole layer is auditable from the chain alone — you do not have to take our word for where the pointer was.

2 · THE SPAWN CYCLE

The formation's starting height steps down on a fixed nine-wave cycle and then repeats from wave 10. It never starts below Y=64. The cap is deliberate: difficulty plateaus, so a run is an endurance test with a knowable ceiling rather than an escalation you already know you cannot win.

WAVE	1	2	3	4	5	6	7	8	9	10 →
SPAWN Y	120	80	72	72	72	64	64	64	64	repeats from 120

Bunkers are never rebuilt between waves. Whatever you eroded in wave 1 is still eroded in wave 6, while the rack starts 56 pixels lower. That is the entire difficulty curve, and it has no numbers in it that you did not create yourself.

3 · THE ALIEN RELOAD TABLE

How fast the aliens shoot back is indexed by your score, not by the wave number and not by a clock. A good player reaches their own cliff earlier because they earned it.

SCORE REACHED	RELOAD DELAY	RELATIVE DENSITY
0 – 511	48	×1.00
512 – 4,095	16	×3.00
4,096 – 8,191	11	×4.36
8,192 – 12,287	8	×6.00
12,288 and above	7	×6.86

There are three alien shot types. Only one of them — the Rolling shot — tracks your column; the other two are pattern noise. Shots step 4 pixels normally and 5 pixels once eight or fewer aliens remain.

THE MARCH

Rack step interval = clamp(917 ms × alive / 55, 33 ms, 1100 ms), driven by a wall-clock accumulator and never by a frame counter — so a 144 Hz monitor plays the exact same difficulty as a 60 Hz one. Your own kills are what accelerate the thing that is killing you. The rack steps 2 pixels (the last alien 3 right, 2 left) and drops 8 pixels on the step after it touches an edge. You lose when the rack reaches the cannon's row, not the bunker line.

4 · SCORING

TARGET	POINTS	NOTE
Bottom two ranks (22 aliens)	10	closest, fastest to farm
Middle two ranks (22 aliens)	20
Top rank (11 aliens)	30	furthest, slowest — the one bullet is the price
Saucer	50 – 300	from the published table above
Maximum per rack	990	(22×10) + (22×20) + (11×30)

Extra base: exactly one per run, at 1,000 points (operator switch: 1,000 or 1,500). There is no repeating extra-life threshold.

Clean-wave streak: clear a wave without any bunker dropping below 90% integrity and the multiplier climbs ×1.0 → 1.1 → 1.2 → 1.3 → 1.4 → ×1.5. The first breach resets it. It is off in PURE '78.

5 · THE TAPE

Every payload broadcast by a run carries the same header, then a body that depends on the kind:

OP_FALSE OP_RETURN <push N>

  'MINV'   4 bytes, ASCII   the app tag
  ver      1 byte           0x01
  kind     1 byte           see the fee table below
  seq      2 bytes, big-endian   the run's monotonic event counter
  body     kind-specific

THE DIGEST

The closing SEAL carries tapeDigest, computed like this:

tapeDigest = sha256( concat( payloads sorted by ASCENDING seq ) )[0 .. 15]
             // 15 bytes = the first 30 hex characters

To verify a run yourself: take the funding txid (the tape root), walk its sub-chains on any block explorer, decode each OP_RETURN, sort by seq, concatenate, hash, and compare against the SEAL. A match is VERIFIED. A missing or reordered event breaks the digest. The replay page does exactly this and nothing more.

The precise scope of the ordering claim. Order comes from the seq field and from the SEAL digest that binds the complete set — not from the spend graph, because a run uses two parallel sub-chains. We are not claiming wall-clock ordering across chains, and we will not pretend to.

And the anti-cheat claim, stated honestly. None of this makes botting impossible. What it does is make a faked score cost exactly what a real one costs: roughly 800 sequentially dependent transactions, broadcast at human cadence, against real block space. The rate is the proof. You will not find a "cheat-proof" badge anywhere in this build, because it would be a lie.

6 · WHAT EACH TRANSACTION COSTS

Fee formula, identical for every kind: max(25, ceil((209 + payloadLength) × 0.12)). You pay the miner. The house takes nothing out of these.

KIND	BYTE	WHEN	BODY	SATS
SHOT	0x01	you pull the trigger	11 B	27
RESOLVE	0x02	that bullet resolves	13 B	27
UFO	0x03	the saucer is destroyed	16 B	27
WAVE	0x04	the last alien of a rack dies	24 B	28
DEATH	0x05	a base is lost	12 B	27
CONFIRM	0x06	a real block landed — off by default	15 B	27
ECHO	0x07	THE CHRONICLE's sweep — perk only, batched	16 B	27
TURRET	0x08	THE MINTER's turrets — perk only, batched	16 B	27
WING	0x09	THE CYBORG's wingman — perk only, batched	16 B	27
SEAL	0x0A	the closing record	40 B	30
SPLIT	—	fan the funding into 2 sub-chains	—	41
SWEEP	—	return every leftover sat to you	—	120

WHOLE-RUN ARITHMETIC

RUN	INTERACTIONS	SATS	SUSTAINED
Short — 3 waves, 250 shots	509	13,910	~2.8 tx/s
Median — 5 waves, 400 shots	814	22,147	~3.0 tx/s
Long — 8 waves, 620 shots	1,259	34,192	~3.2 tx/s

A default session funds 250,000 sats — about eleven median runs from one wallet popup. Whatever you do not spend is swept back to your own address when the run ends, and a session abandoned by a page reload is swept on your next visit. Nothing is ever stranded.

THE HOUSE CUT

Session rake: 0 basis points. Every money feature in this build ships switched off, and the rake is one of them. If it is ever turned on, the split becomes a visible second output in your own funding transaction — auditable on WhatsOnChain, per transaction — and the exact number replaces this paragraph on this page before it goes live, not after.

7 · THE METERING RULES

These exist so the game can never flood the relay, and they are hard limits in code, not guidelines:

At most 2 transactions per frame. A third is deferred to the next frame — the alien plays its explosion but is not removed until its emission is reserved, so the deferral is free to look at.
A global 8 transactions per second token bucket. Refusing a shot shows MEMPOOL FULL.
Two parallel sub-chains, each sleeping 45 ms after a successful broadcast. Throughput comes from parallel chains, never from removing the throttle.
The fire governor: you cannot fire while more than 3 events are queued. The depth is on screen at all times, so a blocked trigger is never mysterious.
There is no screen-clear primitive anywhere in this build. No smart bomb, no nuke, no column laser. Pierce is hard-clamped at 2. Every composed ability is a metered sweep capped at one removal per 60 ms.

The one event that removes many aliens at once costs nothing to broadcast. When a real block confirms, the bottom rank lifts off the rack. Those aliens are not kills: they pay zero points and emit zero transactions, in every configuration, because the mempool drained itself and you did not shoot. That is the structural safety guarantee of the whole design.

8 · WHAT RANKED REFUSES

Ships are cosmetic in RANKED, and it is enforced in four independent places rather than promised in one:

Perk resolution returns an empty set whenever the mode says so, whatever hull is selected.
Casual-only hulls are refused outright and fall back to the stock cannon.
The server rejects any submission whose perk hash is not the hash of the empty set, or whose seed is not today's daily seed.
The strongest one: autonomous perks emit their own transaction kinds. An ECHO can only exist if THE CHRONICLE was flying. The verifier rejects any ranked tape containing a perk-only kind — so fairness is not a promise, it is an audit anyone can run.