The usual argument about whether slaves are worth keeping is conducted in the wrong currency. The wiki’s Slavery entry covers the social side well — slaves run on Submission instead of Loyalty, and low Submission produces an uprising rather than a strike — but it does not publish a guard ratio, a food cost, or an attrition rate, because those are things you set and measure in your own save, not constants the game prints anywhere. Answer the labour question first and the social question gets easier to reason about.
Worked example, not a game constant: with this calculator’s illustrative default inputs — 500 slaves at a 0.6 output ratio, one guard per 8 slaves, and 20% yearly attrition — the break-even output ratio comes out to about 0.29. Below that ratio the arrangement is a net labour loss no matter how many slaves you add; at the assumed 0.6 it nets roughly 156 worker-equivalents a day, with guards and food overhead each eating about 62 of that. Your own save will not match these numbers — enter your own below.
Songs of Syx slave ledger, at a glance
| Situation | Net labour | What to do about it |
|---|---|---|
| Too few guards per slave, low output ratio | -187 net worker-equivalents/day | This arrangement is a net loss of labour — either raise the output ratio or thin out the guard-to-slave ratio before adding more slaves. |
| Small, well-guarded operation under the unrest line | 156 net worker-equivalents/day | This is sustainable at current scale — the margin is thin enough that a bump in attrition or guard duty is worth re-checking. |
| Large slave share, output pays off but past your own risk line | 319 net worker-equivalents/day | The labour math is fine, but growth from here buys output at the cost of stability — that trade is worth making deliberately, not by accident. |
These are outputs of this site's own worker-equivalent accounting run against illustrative inputs we chose, not values extracted from the game files — output ratio, guard coverage and attrition are things you measure on your own save. Use the calculator below for your own numbers. Replacement work-days were scaled down with the 16-day year default so the three rows keep the same bad, marginal and good shapes.
The question is an accounting question
The argument usually runs on whether slaves are good, which is unanswerable, when the tractable question is whether they are net positive in labour. Slaves do not arrive free of overhead. Before a single unit of anything is produced, an arrangement of any size is already paying:
- Guards, who are workers you have taken out of production.
- Food, which is itself the output of farmers you also had to employ.
- Replacements, because losses have to be made good by raids or purchases that cost labour or money you earned with labour.
Denominate all four in the same unit — worker-equivalents per day — and the argument collapses into arithmetic. The answer is not universal. It swings on your output ratio, your guard coverage and how expensive acquisition is where you are on the map, all of which differ per save.
Price your own arrangement
Slave ledger: what the arrangement actually nets you
156 worker-equivalents net per day 300 gross, less 144 in guards, food and replacements
- Slave share of population
- 20%
- Guards tied up
- 62.5
- Food cost in farmers
- 62.5
- Replacing losses costs
- 18.8 / day
- Net per slave
- 0.31
- Break-even output ratio
- 0.29
Net 156.3 worker-equivalents a day, 0.31 per slave. Break-even output ratio is 0.29 against your 0.60, so you have 109% of headroom before this stops paying.
Output ratio, guard coverage and attrition are yours to measure — pause on your own population screen and read them. The defaults exist to show the shape of the sum, not to claim your version's numbers. Everything is denominated in worker-equivalents so guards, food and replacements can be subtracted from the same total.
The output ratio, the guard coverage and the acquisition cost are inputs because they are things you can read and we cannot. We are not going to assert a number for how hard a slave works in your build; the game has been reworking its law, order and slavery layers and the community documentation for those layers does not exist yet. What does not change between versions is the shape of the sum, and that is what this is.
The row to read first is break-even output ratio. It is the productivity level at which the whole apparatus exactly pays for itself. Compare it to the ratio you entered. The distance between the two is your entire margin, and it is usually much narrower than the debate suggests.
Three results that surprise people
Run a few configurations and the same three patterns fall out, and none of them are things a prose guide tends to mention:
- Guard coverage dominates at scale. Because guards are a fixed fraction of slaves, the cost grows exactly as fast as the output. Doubling the arrangement does not dilute the guard bill; it doubles it. There are no economies of scale in the deduction column, which is why “just get more slaves” stops working at a size that varies by save.
- Food is not free even when you have plenty. Converting slave rations into the farmers who grew them is the step everyone leaves out, and at a low farm yield it can exceed the guard bill. This is also why a food-constrained city gains less from slavery than a food-rich one, independent of anything social.
- Attrition is the quiet one. A modest yearly loss rate with an expensive acquisition path produces a standing daily cost that never appears on any panel, because it is paid in expeditions and coin rather than in a line item.
Where the labour maths stops being the whole answer
The ledger deliberately stops at labour. It does not price unrest, and it should not pretend to, because the consequence of unrest is not a rate — it is an event. The wiki’s slavery entry documents that slaves run on their own Submission stat rather than the citizen Loyalty stat, that submission is lower when there are too many slaves relative to citizens and guards, and that low-submission slaves attempt a violent uprising against the throne instead of the strikes or riots unfulfilled citizens produce. What the calculator does give you is the population share, checked against a threshold you set yourself.
That threshold is a judgement, not a fact, and the calculator treats it as one. Set it where your own experience puts it. What the tool then guarantees is only this: if you cross your own line, you will be told, in the same view where you can see exactly how much labour the crossing bought. That trade is the actual decision. Everything else in the debate is preamble.
If the arrangement clears break-even but your city is still unhappy, the cause is somewhere else entirely — start from which need is capping the rest rather than adjusting the labour force, and check the food side in why your days of food never go up, since slaves are eaters your demand figure may not be counting.
V71 slavery rework, and why the ledger has no baked-in figure
V0.71 — Reign of Terror, released 2026-06-15 and the branch that reworked slavery from the ground up.
We are not going to give you a V0.71 figure for slave productivity or guard requirements, because we have not measured them and the sources that publish such figures do not say which version they measured. The ledger asks you for them instead. That is a deliberate trade: less to copy, nothing to go stale.
Where another source does publish a slavery figure, check what version it was written against before you use it. The most-linked public wiki pages for this subject predate the current social systems by years. That does not make their numbers wrong — it makes them untested, which is a different thing and worth keeping separate in your head.
Sources
- Songs of Syx Wiki, Slavery — the Submission stat, its link to guard presence and slave-to-citizen ratio, and uprisings as the slave-specific unrest event; the page documents the mechanic but publishes no guard, food or attrition figures
- V71 - Reign of Terror - Available Now (Steam) — confirms slavery was reworked from the ground up in V71, including slaves becoming a fully tradeable good, and that no later major version has posted since
