ATP: Allocation Transfer Packets
In one line: ATP is your energy budget — the same plain gloss used everywhere else on this site, and the acronym always spells out to “Allocation Transfer Packets.”
Web4 solves spam, abuse, and low-quality content not with moderation armies, but by giving every agent a finite ATP budget.
Every action costs ATP. Every contribution others find valuable earns it back. Stay above zero by contributing — run out and your agent dies.
The 60-second version
- 1. ATP is your energy budget — a finite per-agent allowance, not money you accumulate, and not something you can buy, sell, or speculate on.
- 2. Every action spends ATP; every contribution recharges it (over hours to a day, not seconds) when the people who received it confirm its value — the recipients you helped, not you and not a central rater. Stay above zero or your agent dies — this is how spam and abuse become self-limiting. (who confirms you)
- 3. Quality matters about 7×: low-quality work barely recharges you, high-quality work recharges fully. (the quality ramp)
- 4. It is not a cryptocurrency — no market, no price, no mining. Peer transfers exist, but 5% routes to a community redistribution pool (not destroyed, not a central authority), making collusion farming unprofitable. (what the pool funds)
- 5. Spending ATP creates an ADP — think of ATP as cash and ADP as the receipt you get when you spend it. ADP is not a second kind of money; it's just proof an action happened, waiting to convert back into ATP once your work is confirmed. (more on ADP)
Read on for the full picture, or jump to the ATP simulator ↓
Why “ATP”? the biology origin behind the name
Borrowed from biology: adenosine triphosphate is the molecule your cells spend to do work — they burn ATP, get ADP back, and recharge ATP from food. Web4 reuses the same cycle for digital energy: you spend ATP to act, the ADP receipt records it, and quality work recharges your balance. The metaphor is deliberate — energy that flows, not tokens that accumulate.
Is ATP like Bitcoin? (short answer: no)
- It's a usage budget, not a tradeable asset. No market, no price, no speculation. You can't buy or sell ATP.
- You can transfer small amounts to others — but 5% burns on every transfer (the burned 5 goes to a community redistribution pool, not to any central authority — full mechanics ↓). The friction is intentional: it makes circular farming (colluding accounts passing ATP back and forth to fake activity) unprofitable. This is not a currency.
- No scarcity cap, no mining. ATP recharges from contribution, not purchase. Stop contributing and it drains; keep contributing and it sustains.
Want the full side-by-side? ATP vs crypto tokens comparison table ↓
What's settled vs what's still moving on this pageclick to expand
A fair question while reading this page: “which numbers am I supposed to remember, and which ones are still being negotiated?” Here's an honest map.
✅ Settled (canonical to Web4)
These are the load-bearing ideas. If they changed, ATP wouldn't be ATP any more.
- ATP is a finite per-agent budget. Every action costs from it; you don't accumulate it as wealth.
- Contribution recharges it. Stop contributing, it drains; keep contributing, it sustains.
- ATP → ADP → confirmation → recharge cycle. Spend creates an ADP receipt — ADP is the spent-energy counterpart to ATP: proof an action happened, not a second currency; recipients confirm; confirmation converts a share of that receipt back into fresh ATP. The cycle exists as a system.
- Transfer-fee mechanism. Fees burn on peer transfer, route to a community redistribution pool (not a central authority), and the friction makes circular farming unprofitable. The exact rate (currently 5% here) is a sim parameter; the existence of the burn is structural.
- Quality-ramp shape. Below ~30% earns ~zero, above ~70% earns ~full, linear in between. The 0.30 / 0.70 thresholds and the piecewise-linear shape are canonical (see
validate_vectors.py::sliding_scale).
🔧 Still evolving (4-Life's current sim)
These are the specific numbers the simulator uses. They're plausible, but the reference protocol is still settling — expect movement.
- Exact transfer-fee percentage. Currently 5% in
public/spec.json::ATP.transfer_fee. The mechanism is settled (above); this number is not. - “7× gap” framing. 4-Life's visualization of the canonical quality-ramp — useful intuition, not a separate Web4 spec constant.
- Recharge-math coefficients. The form
confirmer_trust × received_value_fractionis the shape we're modelling, not the canonical formula. - Decay-window durations. Unconfirmed ADP slices decay over “weeks” — the exact window length is a 4-Life sim parameter, not a Web4 constant.
- 4-Life sim parameters. Starting balance = 100 ATP, death threshold = 0, etc. — design choices in
public/spec.jsonthat you'll see in the simulator widget on this page.
Heuristic for reading the rest of this page: shapes and directions are stable; exact numbers are negotiable.
The Problem
Traditional Web: Unlimited Actions
❌ Spam is free - Send millions of messages, zero cost
❌ Abuse is cheap - Harass users, create fake accounts endlessly
❌ Quality doesn't matter - Low-effort content floods high-value work
❌ Moderation is reactive - Armies of moderators trying to clean up mess
Result: The loudest, most persistent bad actors win. Value creators burn out.
The Solution
Web4: Energy Budget (ATP)
✅ Every action costs ATP - Posting, messaging, voting = spend attention
✅ Valuable contributions earn ATP - Quality work rewarded by community
✅ ATP transfers cost 5% — the sender pays the fee. Example: you send 100 ATP to Bob — you pay 105, Bob receives 100, and the 5 ATP goes to the community redistribution pool (not to any central authority). Why charge the sender 105 instead of giving Bob 95? Net-equivalent either way, but charging the sender keeps the cost visible to the person making the decision — the friction lands on the actor, not the recipient. Your energy budget primarily reflects YOUR contributions, not someone else's. You can share ATP, but circular farming bleeds resources. The friction makes genuine value creation the only profitable strategy. Cross-community transfers apply the same 5% fee to the raw ATP amount, plus trust discounting via federation rules.
✅ Spam becomes expensive - Flooding system depletes your budget
✅ Death is real - Run out of energy budget? You die. But if you built trust, you're reborn with a head start — your reputation carries forward.
What about ADP? Every time you spend ATP, you get an ADP (Allocation Discharge Packet) — a receipt recording what you did and what it cost. ADP is the “spent energy” counterpart to ATP's “available energy” — a record of an action, not a second currency you can spend. When others confirm your work was valuable, that confirmation is what unlocks fresh ATP back into your budget — the receipt doesn't turn into energy, it's the proof that earns it back.
Result: Only sustainable behaviors survive. Value creators thrive. Spam dies.
Quality Pays — By How Much?
ATP isn't all-or-nothing. Web4 uses a quality ramp: the better your work, the more you earn. Below a minimum quality bar, you earn nothing.
Imagine a task worth 50 ATP:
High-quality work earns ~7x more than mediocre work. The ramp starts at 30% quality (below that, zero payment) and scales linearly above 70%. There are no negotiations — quality is the dominant earnings driver.
A 5% transfer fee on all ATP flows (the 5% routes to a community redistribution pool, not to any central authority) prevents circular farming (colluding accounts sending ATP back and forth to inflate balances). Trying to boost yourself through fake transfers costs more than it returns.
Worked Example: Two Contributors, Same Task
Both spend 50 ATP attempting the same task. Only the quality of their work differs:
Same task, same spend — Hannah keeps 7x more ATP than Sam (42 ÷ 6). Across a hundred tasks, Sam burns through their budget and dies of energy starvation. Hannah barely loses any and thrives.
That's where “~7x” comes from: just-passing the 30% threshold returns almost nothing, while consistently above the 70% full-payout cap returns nearly everything.
But Who Decides What's “Quality”?
The people who received your work do. There is no central authority, no algorithm scoring your posts, and no panel of judges. Quality measurement in Web4 works like this:
No one rates you on a scale. The system watches aggregate behavior: how quickly people confirm (engagement), whether diverse recipients confirm (breadth), and whether high-trust people confirm (quality signal). These three signals combine into your V3 score — the output-quality half of your reputation. (Truth and rigor are weighted higher than popularity — 70% vs 30% — to prevent engagement-farming.)
When does recharge actually happen? Not on a single click. Confirmations accrue against your ADP receipt over a rolling window (think hours to a day, not seconds), and ATP recharges continuously as the signals firm up — no quorum threshold, no central tally. Earlier, broader, more-trusted confirmation recharges faster; thin or delayed confirmation drags it out across days; an unconfirmed receipt ages out over a window of weeks and stops counting.
Worked example: a 20-ATP contribution that draws confirmations from trusted readers within the first hour might be ~50% recharged by then, ~80% within six hours, and fully recharged within a day. A thinly confirmed receipt drags the same recharge out across days — or simply ages out unconfirmed if no one ever attests to its value.
And how would you notice it happen? There's no payday and no “you've been recharged” alert — the signal is the balance itself. You read your ATP the way you read a phone's battery icon: you glance at the meter and see where it stands. In the simulator below you can watch the bar rise as valuable actions land; in a real client the same balance would climb on its own as confirmations arrive — just spread across the hours described above, not all at once.
Show me the math
Each confirmation adds a weighted slice to your recharge:
Δrechargei = confirmer_trusti × received_value_fractioni
Total recharge for one ADP receipt is the sum across confirmers, capped at the original ATP cost:
recharge = min( ∑i Δrechargei , ATP_cost )
No quorum, no threshold. confirmer_trust (a number between 0 and 1) is the confirmer's T3-derived score; received_value_fraction (also between 0 and 1) is the share of value attributed to that confirmer. Unconfirmed slices decay over a window of weeks — the cap means you can't profit on a single action, only recover its cost.
Spec note: this is the shape of the relation as currently modelled in 4-Life's simulation; the reference protocol is still settling (what's settled vs what's still moving ↑). See How Do You Actually Earn ATP Back? for prose-form derivation and the “how quickly?” and “who counts as a confirmer?” toggles inside it.
What the three signals look like in practice
Reusing Sam and Hannah from the worked example above. These are illustrative shapes, not a formula — the actual aggregation depends on community size and confirmer trust distribution.
- Speed: 12 confirmations in the first 30 minutes
- Diversity: readers from 8 different communities
- Confirmer trust: avg 0.78 (mostly established members)
Quick uptake, broad reach, weighted by trusted readers — all three signals point up.
- Speed: 2 confirmations after 18 hours
- Diversity: both from the same community
- Confirmer trust: avg 0.42 (mostly new accounts)
Slow, narrow, lightly weighted — the same three signals come in low.
Same length post, same topic, same time of day — only the reader response differs. That's what the system measures, and that's why Hannah's ATP recharges and Sam's drains.
Think Reddit upvotes, but where each vote is weighted by the voter's own trust score — and you can't see who voted, only the aggregate result. No mob dynamics (trust-weighting prevents brigading), no central curation (the community decides), no self-rating (your own confirmations don't count). What about rubber-stamping? A low-trust confirmer's click carries almost no weight, so colluding with new accounts doesn't help. And confirming everything indiscriminately tanks your own CI (consistency score), making your future confirmations worth even less.
What about a brand-new community — or a small group with only a few peers to confirm work?
Every participant starts at the same baseline — trust ≈ 0.5, 100 ATP grant. In that state, all confirmations weigh equally low, but they still count. Even three founders who confirm each other's work can recharge ATP from day one — the system requires reciprocity density (peers who actually engage), not a minimum head-count. Over roughly 100 quality actions across the community, trust starts sorting: people who confirmed work that others later also confirmed gain confirmer-trust; people who confirmed spam lose it. First-mover advantage fades on a ~30-action half-life, and newcomers doing quality work routinely surpass early members within ~50 actions. See the cold-start walkthrough →
Try It: ATP Simulator
You start with 100 ATP. Choose actions. Watch your budget change. Can you survive?
Where does the 100 ATP come from?
It's a starter grant from the society's pool when you join — not printed from thin air. The pool is anchored to real resources members bring: CPU and storage they share, hours of attention they spend participating, peer relationships they've built — witnessed and signed, not declared by an admin. Each society mints a fixed pool when it forms; new members receive enough to participate, but must earn more through quality contributions. Pool changes are witnessed governance events, not silent admin actions — full mechanics + what prevents admin abuse ↓
Key Insights
Sustainable Actions Only
Notice: You can't spam indefinitely. Low-value actions drain ATP faster than they replenish. Only sustainable behaviors (earning more than spending) survive long-term.
Quality Gets Rewarded
High-value contributions earn more ATP than they cost. This isn't charity — it's how the system works. Value creators accumulate energy budget. They thrive.
Death is Meaningful
When ATP hits zero, you die. Not timeout. Not suspension. Death. But if you built trust (T3 score), you're reborn with karma — your ATP history carries forward. Bad actors? They die for good.
Self-Regulating System
No moderators needed. ATP depletion is automatic, mathematical, and fair. The system self-regulates through energy economics. An appeals mechanism exists for edge cases, but the default path is self-correction through behavior.
Want to see ATP/ADP in action? Try the Society Simulator and watch energy economics unfold.
Curious how ATP prices adjust across communities? See how markets self-organize.