Space · Budget Engine

Attention Budget Boost

Every widget on a user's Space is automatically scored for attention cost (0.0–3.5). Widgets above 3.0 are blocked as “unsafe”. A boost lets you pay to raise your effective threshold — but the pricing is doubly exponential: it scales with both the boost amount and your install base. A small indie dev can afford a slight boost; a large provider trying to dominate every Space pays astronomically.

Concept

How it works

Default threshold: 3.0

The Space Budget Engine scores every widget across 7 dimensions (complexity, content volume, interaction load, live pressure, volatility, social pressure, historical cost). Any widget scoring above 3.0 is blocked from auto-placement.

Boost raises the gate

Purchase a boost of +0.5 and your widgets are safe up to score 3.5. Purchase +1.5 and they're safe up to 4.5. The true attention cost is still reported accurately — only the installation gate is relaxed.

Surface cap stays absolute

The 14.0 total surface attention budget is not boostable. A boosted widget still consumes its full attention cost against the surface budget. One provider cannot monopolize an entire Space.

Capped at +2.0

Maximum boost is 2.0 points (effective threshold 5.0). Minimum is 0.1 in 0.1 increments. Widgets above 5.0 are considered genuinely too heavy for any Space.

Pricing

Doubly exponential formula

The fee is exponential in both the boost amount and the square root of your install count. This makes small boosts affordable for indie developers while making attention dominance at scale prohibitively expensive.

fee = BASE_RATE × (e^{K_b × boost} − 1) × e^{K_u × √installs}

BASE_RATE

$5.00

K_b

3.0 (budget exponent)

K_u

0.025 (userbase exponent)

Why doubly exponential? A linear or even singly exponential fee would still let large providers buy their way to dominance. The double exponential creates a natural ceiling: the fee for a +2.0 boost at 100k installs is ~$585k/month. No rational actor would pay that, which is exactly the point — it makes attention hogging economically irrational.

Reference

Pricing examples

Representative monthly fees at various boost levels and install counts. Your actual fee is calculated in real-time based on your current install count.

Boost	Threshold	50 installs	500 installs	5,000 installs	50,000 installs
+0.3	3.3	~$9	~$12	~$25	~$100
+0.5	3.5	~$17	~$30	~$66	~$258
+1.0	4.0	~$100	~$210	~$559	~$2,180
+1.5	4.5	~$465	~$1,445	~$2,607	~$10,174
+2.0	5.0	~$2,100	~$10,500	~$24,513	~$538,814

Fees recalculate daily based on actual install count. If your installs drop, your fee drops. If installs reach zero, the boost is auto-cancelled.

API

Boost endpoints

All boost endpoints are scoped under the authenticated developer prefix. You must own the package to purchase or modify a boost.

Method	Path	Purpose
GET	/v1/providers/packages/:packageId/boost	Get active boost for a package (returns `null` if none)
GET	/v1/providers/packages/:packageId/boost/preview?boost_points=1.0	Preview fee for a given boost level without purchasing
GET	/v1/providers/packages/:packageId/boost/tiers	Full pricing table (20 tiers, 0.1–2.0) for current install count
POST	/v1/providers/packages/:packageId/boost	Purchase or update boost (`{ "boost_points": 1.0 }`)
DEL	/v1/providers/packages/:packageId/boost	Cancel active boost
GET	/v1/providers/packages/:packageId/boost/history	Fee event audit trail with pagination

Examples

Purchase a boost

Request

// POST /v1/providers/packages/:packageId/boost
{
  "boost_points": 1.0
}

Response

{
  "boost": {
    "id": "b7a1c2d3-...",
    "package_id": "a1b2c3d4-...",
    "boost_points": 1.0,
    "status": "active",
    "current_fee_cents": 19700,
    "fee_formatted": "$197.00",
    "install_count_at_price": 1000,
    "effective_threshold": 4.0,
    "effective_from": "2026-04-04T12:00:00Z"
  }
}

Preview fee before purchasing

// GET /v1/providers/packages/:packageId/boost/preview?boost_points=1.5

{
  "fee_cents": 307600,
  "fee_formatted": "$3,076.00",
  "effective_threshold": 4.5,
  "install_count": 5000,
  "breakdown": {
    "budget_multiplier": 89.017,
    "userbase_multiplier": 5.754
  }
}

Get pricing table

// GET /v1/providers/packages/:packageId/boost/tiers

{
  "tiers": [
    { "boost_points": 0.1, "fee_formatted": "$4.59", "effective_threshold": 3.1 },
    { "boost_points": 0.2, "fee_formatted": "$6.33", "effective_threshold": 3.2 },
    // ... 18 more tiers ...
    { "boost_points": 2.0, "fee_formatted": "$37,740.00", "effective_threshold": 5.0 }
  ],
  "install_count": 5000
}

Billing

Monthly fee recalculation

Your boost fee is recalculated daily based on your current install count. Changes are applied when the delta exceeds 5% or $10 (whichever comes first).

Installs increase

As more users install your package, the userbase multiplier grows exponentially (e^{0.025 × √installs}). Your monthly fee increases accordingly. A fee event is logged for audit.

Installs decrease

If users uninstall, the fee decreases. If install count drops to zero, the boost is auto-cancelled. You can always repurchase later.

Full audit trail

Every fee change is logged in the provider_boost_fee_events table with the full formula inputs (base rate, exponents, boost, installs, raw result). Use the /history endpoint to review.

Transparency guarantee: The pricing formula and all constants are documented here and are the same for every provider. There is no hidden pricing, no enterprise negotiation, and no special deals. The math is the same whether you have 10 installs or 100,000.

Design

Why this model?

Calm Spaces by default

Without a boost, the 3.0 threshold keeps Spaces calm and balanced. Heavy widgets must earn their place through explicit, expensive opt-in.

Indie-friendly

A small developer with 50 installs pays ~$22/mo for +0.5 boost. The exponential curve only becomes punishing at scale, where the provider is already extracting significant value.

No attention monopoly

The surface cap (14.0 total) stays absolute. Even with a max boost, a single provider's widget still competes for surface budget against all other widgets. Boost relaxes the per-widget gate, not the per-surface cap.

True cost is visible

The actual attention_cost is always reported accurately to the user. Boost doesn't hide or reduce the score — it only changes whether the widget is allowed to be installed.

Quick Reference

Limits and constants

Parameter	Value	Notes
Minimum boost	0.1	Must be a multiple of 0.1
Maximum boost	2.0	Hard product limit
Default threshold	3.0	Unchanged — applies to all packages without a boost
Max effective threshold	5.0	3.0 default + 2.0 max boost
Surface cap	14.0	Not boostable — absolute per-Space limit
Base rate	$5.00	BASE_RATE_CENTS = 500
Budget exponent (K_b)	3.0	Controls boost cost curve steepness
Userbase exponent (K_u)	0.025	Controls install count cost scaling
Recalculation frequency	Daily	Fee adjusts when delta > 5% or > $10
One boost per package	Yes	Purchasing a new boost replaces the previous one