Skills

Status: draft v0.1, 2026-11-09. Hangs off GAME-DESIGN.md §5 (the player arc) and is the backbone of horizontal progression. Numbers are shapes, not balance.

---

1. Training happens in real time, whether you play or not

Borrowed from EVE, because EVE got this right and nothing else comes close for a game like ours.

• You queue a skill. It trains against the wall clock, continuously, offline, forever.
• There is no grinding, no XP, no repeating an activity to get better at it.
• A player who is away for a week comes back further along, not behind.

This is the single best fit between a progression system and our anti-pillars. No daily-login jail, no session-length arms race, no punishing people with jobs and children. The queue is also perfect check-in material: thirty seconds to see what finished and line up the next thing.

Skills are never lost. Not on death, not on bankruptcy, not on losing a war.

2. Levels are infinite. The benefit is not.

There is no level cap. A skill can always be trained one more time, forever. But:

• Time per level grows exponentially. Each level takes noticeably longer than the last.
• Benefit per level shrinks geometrically. Each level is worth less than the last.

Which means the total benefit of a skill converges. The sum of an infinite series of shrinking gains is a finite number, and that number is the real ceiling — approached forever, never reached.

The asymptote is the cap. There is no wall to hit, and no one ever runs out of things to train, but nobody ever becomes categorically better than a focused newcomer either.

2.1 The curve

Mastery held after t minutes of training:

mastery(t) = 1 − exp(−a · t^b)      a ≈ 0.021, b ≈ 0.356

A stretched exponential: quick at first, then an ever-flatter approach to a ceiling it never reaches. Two constants, and they produce this (measured, npm run sim:skills):

Trained for	Mastery
10 minutes	4.6%
1 hour	8.8%
1 day	24.2%
1 week	41.7%
2 weeks	50.4%
1 month	58.9%
3 months	72.2%
6 months	82.8%
1 year	89.4%
3 years	95.0%
10 years	99.4%

Useful in minutes, competent in a day, genuinely specialised in a fortnight, mastered in six months, and improving forever after at a rate nobody would organise their life around.

The six-month figure is 7 points under target (82.8% against 90%). Two constants cannot bend enough to hit all five anchors at once; this is the best honest fit and the miss is recorded rather than hidden. A third parameter would close it, at the cost of a curve nobody can hold in their head.

2.2 Why this produces breadth without forbidding depth

We do not need a rule against over-specialising. The curve does it:

• Once you are deep in a domain, the next level there costs weeks and gives almost nothing.
• The first levels of a new domain cost minutes and give a great deal.
• So any rational veteran spreads out. Breadth becomes the obvious play rather than an imposed one, which is always better design than a cap.

A five-year veteran is therefore not a better miner than a two-week specialist in any way that decides an outcome. They are a miner and a scanner and a hauler and a fabricator, able to switch roles as the galaxy changes. Versatility is a genuine reward and it costs the newcomer nothing.

The newcomer's counter-play is focus: pick one thing, spend a fortnight, and stand level with anyone alive at that one thing. Then go be meticulous and well-informed, which is where outcomes are actually decided (COMBAT.md §10).

2.3 The game stops demanding your attention as you age into it

A deliberate and slightly unusual consequence, and we should protect it.

Early on, training moves fast and checking in is rewarding. Later, a single level takes weeks — so a veteran can disappear for a month and lose nothing meaningful. The game asks less of you the longer you have played it.

That is the opposite of how most live games treat their oldest players, and it is correct for ours: people who have been here for years are the ones with the least time to spare. A game they can step away from without penalty is a game they come back to.

3. What a skill does

Two jobs, in order of importance.

3.1 Access — what you may use at all

The primary role. A skill gates hulls, modules, structures and activities. Without the skill, the equipment is simply not usable.

This is how progression stays horizontal: skills open doors, and the number of doors you can walk through is what grows forever. Having access to a deep-survey array, a heavy hauler and an industrial refinery is a meaningfully larger life than having access to one of them — without any of them being stronger than they were.

3.2 Efficiency — how well the equipment performs in your hands

The second role, and not a small one.

Effective performance = what the equipment can do × what you know about doing it.

Mastery is a position on a curve; the curve maps onto effect. The 0–100% mastery range becomes a multiplier on what the equipment does, from 1.00× to 4.00×.

The band is wide on purpose. A decade of training has to be felt, or the long curve is a cosmetic.

	Mastery	Effect
Fresh pilot	0%	1.00×
One day	24.2%	1.73×
Two weeks	50.4%	2.51×
10 years, spread over 10 domains	89.6%	3.69×
10 years, one domain	99.4%	3.98×

Read the shape, not the endpoints. Most of the band is bought in the first fortnight: one day of training already doubles a fresh pilot, two weeks reaches 63% of the whole range, and the remaining third costs years. A decade-old veteran is 58% better than a two-week specialist — decisive in a duel, recoverable with a second ship, better intel, or a better rock.

Skills never stack. One skill applies to one activity, and two skills never multiply together. Without that rule a 4× ceiling compounds to 16× and the numbers stop meaning anything.

This band is the most dangerous constant in the game. It decides how much of a fight is settled before anyone undocks, and COMBAT.md has to be designed knowing a 4× pilot exists.

3.3 What a veteran actually looks like

Nobody spends ten years on one domain. There are many, and mastering them all is not a plan anyone would make. The real question is not depth against a newcomer, it is how years get divided:

Career	Domains	Mastery in each	Effect
2 weeks, focused	1	50.8%	2.52×
3 months, focused	1	74.7%	3.24×
1 year, 2 domains	2	82.9%	3.49×
5 years, 6 domains	6	88.0%	3.64×
10 years, 10 domains	10	89.6%	3.69×
10 years, 1 domain	1	99.4%	3.98×

A decade spread across ten domains leaves you roughly where a focused player gets in a year — in ten different things at once. Depth is cheap to start and ruinous to finish; breadth is what years are actually for. The last row is a thought experiment, not a player.

4. Shape of the system

• Infinite levels, each costing exponentially more time (×1.05) and buying a smaller slice of a converging curve (§2). The first level takes two minutes; level 200 takes 24 days and adds a fifteenth of a percent.
• Prerequisites form a tech tree. Advanced equipment needs supporting fundamentals, which gives the tree shape and makes a specialisation feel like a path rather than a shopping list.
• A queue, long enough to set and forget for days. This is a mobile game; nobody should lose training time because they were asleep when something finished.
• No attributes, no implants, no training accelerators. EVE has them; they are complexity, a treadmill, and the kind of thing that turns into pay-to-win the moment money enters the picture. We are not building that.

5. How skills meet the rest of the game

System	What skill changes
Scanning (SECURITY-AND-SOVEREIGNTY.md §7)	Sweep range and resolution; how much of a system a ping reveals
Gathering (GATHERING.md)	Which deposits you can detect at all, and extraction rate
Travel (TRAVEL.md)	Drive range per free-burn hop, hull classes you can pilot
Industry (ECONOMY.md)	Refining yield, fabrication speed, which blueprints you can run
Combat (COMBAT.md)	Which hulls and modules you may fit; modest efficiency on what you fit
Command	Fleet size you can lead, drones you can run at once, structures you can operate

Note what is absent: no skill multiplies damage or hitpoints enough to decide a fight. Combat is settled by composition, counters and information (COMBAT.md §10), and skills must not quietly become the fourth thing — or they become the only thing, given enough years.

6. Why this is the right system for this game

• It respects absence. Progress accrues while you live your life, which is the whole promise of an asynchronous game.
• It is honest about time. Nobody can rush it with money or with grinding. Everyone's clock runs at the same speed, forever, and that is a rare kind of fairness.
• It makes specialisation meaningful. A fortnight of focus puts you at 2.5× against a veteran's 3.7×, which makes corporations want members rather than alts — and we have no alts (GAME-DESIGN.md §10).
• It keeps the door open. A player starting in year ten is two thirds of the way up the band after a fortnight, and closes on a decade-old veteran from there slowly but permanently. They will never match that veteran's breadth, and they do not need to.
• It lets veterans leave and come back. By the time a level takes a month, a month away costs nothing (§2.3).

7. Open questions

• [x] How long is mastery, precisely? Settled: minutes for the first levels, competent in days, quite skilled and specialised in 1–2 weeks, domain mastery at ~6 months, then infinite levels of diminishing value. See §2.1.
• [ ] The exact exponents: what ratio for time per level, what ratio for benefit per level, and what total the series converges to? These three numbers are the progression system, and they want simulating before a line of code.
• [ ] Does training speed depend on anything at all, or is everyone's clock identical? Identical is the fairest and the simplest, and I would need a strong reason to break it.
• [ ] Can skills be reallocated? Our one-character rule (GAME-DESIGN.md §10) argues for a hard-but-possible respec so a player is not permanently trapped by an early choice. Possibly a long, expensive retraining rather than an instant reset.
• [x] Is there a total skill cap? No. Levels are infinite; the converging benefit curve does the work a cap would have done, without ever taking anything away from anyone.
• [ ] Do skills gate corporation roles — who may set a gate's policy, spend the treasury, declare war? Likely, and it belongs in CORPS.md.
• [ ] Should the first hour of play grant a few skills outright, so a new pilot is not waiting on a timer to do anything at all? Almost certainly yes.

---

Changelog

• 2026-11-09 — v0.1. Real-time offline training with a queue, borrowed from EVE. The deliberate divergence: depth caps in weeks, breadth never ends, so veterans are versatile rather than stronger and a year-five newcomer reaches parity in their chosen lane. Skills gate access first and improve equipment efficiency second, modestly. No attributes, no implants, no accelerators.
• 2026-11-09 — v0.2. Hard cap replaced by Jonathan's curve: infinite levels, exponential time, geometric decay in benefit, so the total converges and the asymptote is the cap. Milestones fixed — minutes, days, 1–2 weeks specialised, ~6 months mastery, infinite beyond. Two consequences recorded: breadth becomes the rational play without being legislated, and the game asks less of a player the longer they have played.
• 2026-11-10 — v0.3. §3.3 adds the measured breadth-versus-depth profiles: a decade over ten domains lands near 90% in each, so years buy range rather than an unreachable peak, and 50% is read as a competent specialist rather than half a player. Curve refitted to Jonathan's ladder as a stretched exponential (1 − exp(−a·t^b)): 5% at ten minutes, 24% at a day, 50% at a fortnight, 83% at six months, 99% at ten years. Six months lands 7 points under target and the miss is recorded. §3.2 adds the rule that keeps pillar 6 alive under a much steeper curve: mastery is a position, not a multiplier — the full range maps to a 1.00×–1.35× effect band, so a decade beats a fortnight by 49 points of mastery but only 14.6% of performance.
• 2026-11-10 — v0.4. Effect band widened from 1.00×–1.35× to 1.00×–4.00× on Jonathan's call: a decade has to be felt. A fortnight buys 2.51× and a decade 3.98×, so most of the range is earned in the first two weeks and the last third costs years. Adds the no-stacking rule — one skill per activity, skills never multiply — without which the ceiling compounds to 16×. §3.3 adds measured breadth-versus-depth career profiles.