Factorio: Engineering Addiction#

UPS advanced#

Week 11 established that UPS is a finite, shared budget. This week’s question is which entities spend that budget fastest — because count alone doesn’t answer it.

• for assembling-machine-type buildings, the number of machines is what costs UPS, not their speed — modules and beacons that let fewer, faster machines match the same output are a genuine UPS saving, not just a space saving
• inserters are a specific, callable-out cost of their own — cutting the number of inserters, or clocking them so they don’t check every tick, reduces this line directly
• biters and spitters sit in their own “Unit” entity class, separate from buildings — removing them removes that cost outright, which is why a base that’s cleared its perimeter runs measurably lighter than one still fighting off attacks

Pollution UPS#

Pollution isn’t free to simulate either — it’s a cloud tracked per map chunk, updated on its own schedule, and it’s also what decides how often biters show up to cost UPS in the first place.

• each 32×32-tile chunk’s pollution updates every 64 ticks: once a chunk reaches 15.0 pollution it starts spreading outward in all four cardinal directions at 2% per 64 ticks, generating new chunks if it needs to spread into unexplored map
• a building’s actual output isn’t its listed base value — it’s pollution multiplier × energy-usage multiplier × base pollution, so a heavily module-boosted machine can produce well above its own base figure, not just a flat bonus on top of it

Pollution UPS: the biter feedback loop#

• pollution doesn’t just sit there — spawners absorb it to fuel attacks, so a heavier cloud means more frequent, bigger attacks: pollution converting straight into last slide’s biter-related UPS cost
• cutting pollution output (fewer, more efficient machines) is a UPS lever precisely because it starves what spawns the entities costing UPS elsewhere

Pollution by building#

Count and type both mattered for machines two slides ago; pollution is the same story with different numbers, and one genuinely counter- intuitive result: a faster machine tier is not always the dirtier one.

Pollution by building: assemblers and furnaces#

Assembling machine 1 → 2 → 3 gets cleaner per machine as it gets faster — the same “fewer, faster machines” move that saves UPS on the entity-count side also cuts pollution on this side, for the same underlying reason: fewer machines running the multiplier formula above.

Pollution UPS, in practice#

Source: Factorio Wiki, CC BY-NC-SA 3.0.

Those hard edges are the chunk boundaries the simulation actually updates along — the cloud isn’t smooth because the underlying calculation isn’t either.

Pollution, in practice#

Source: Factorio Wiki, CC BY-NC-SA 3.0.

This is the per-building table two slides ago, live — the same breakdown, read off a running base instead of memorised.

What a cloud actually buys the biters#

A spawner doesn’t launch an attack because pollution exists — it launches once it’s absorbed enough to afford one, on a schedule with real numbers behind it, not “eventually.”

• every 64 ticks, a spawner in a chunk above 20 pollution absorbs 20 + 0.01 × chunk pollution; below that threshold it absorbs three times the cost of the most expensive unit it could currently spawn instead — either way, it keeps absorbing until it’s banked enough to afford one
• an attack wave costs pollution per unit added: 4/20/80/400 for small/ medium/big/behemoth biter, 4/12/30/200 for the spitter line — a heavier cloud doesn’t just trigger attacks sooner, it buys a bigger wave per trigger

From banked pollution to an attack#

• once enough is banked, the spawner sends a unit to a rendezvous point; every 1-10 minutes (randomised) the mustered group launches, waiting up to 2 more minutes for stragglers still en route
• this is the mechanism behind “heavier cloud, bigger attacks” from the Pollution UPS slide earlier, not just a qualitative link — the same pollution that costs UPS to simulate is the exact currency spawners spend to cost more UPS later

Tile absorption#

Chunks absorb pollution passively, at a rate set entirely by what’s covering the ground — not something a player action changes.

Paving over ground for belts and floors is a UPS-neutral choice on its own, but it’s also a pollution-absorption choice — a base paved edge to edge is absorbing nothing at the perimeter where absorption would otherwise matter most.

Pollution sinks#

Destroying a nest bumps evolution permanently (Week 5); leaving pollution alone lets it keep spreading. A pollution sink is a third option: wall off a cluster of nests, close enough that your own pollution keeps them docile without destroying them.

• this isn’t about pathfinding cost — the UPS bill for biters comes from the “Unit” class actually moving and fighting, not from sitting idle. A contained cluster that never joins an attack stays idle, so it never enters that cost line
• the trade-off is real estate and risk: the sink sits inside territory fenced off permanently, and a wall failure releases every contained nest at once
• a deliberate alternative to clearing everything (pays evolution) and ignoring everything (pays ever-growing attack size), at the cost of babysitting one more piece of infrastructure

Pollution sinks, in practice#

Source: Factorio Wiki, CC BY-NC-SA 3.0.

Every nest inside that wall is still there — it’s contained, not cleared, which is exactly the point: no evolution paid, no attacks launched.

Reducing base footprint#

Week 2 framed footprint as a build-time and travel-time question. Run the same trade-off through a UPS lens and the answer doesn’t automatically match — a smaller footprint isn’t always a lighter one.

• footprint and entity count are related but not the same: a compact base of many small machines can hold more UPS-costing entities than a sprawling one built from fewer, larger ones
• beacons and modules that shrink a line’s footprint do so by cutting machine count for the same output — the same move that cuts UPS cost, so here both axes point the same way
• a smaller pollution footprint also means a smaller cloud, feeding back into fewer biter attacks — footprint, pollution and UPS are one concern viewed three ways

What causes lower UPS#

Diagnosing a slow base means picking the right cause off this week’s checklist, not naming “too much stuff” — each candidate calls for a different fix.

• entity count: raw numbers — more machines, inserters and belts each add their own small cost; the fix is usually consolidation
• entity type: the same count of a cheaper entity type costs less than an expensive one — fewer, faster assemblers beat many slow ones
• pollution and biter activity: scales with output and spread, not base size — the fix is pollution reduction or a cleared perimeter
• clearing biters doesn’t help an entity-count problem, and consolidating machines doesn’t help a pollution problem

This week’s practical#

Goal: audit a supplied base's UPS optimisation and remaining cost
Test: check it against this week's checklist — entity count, entity
      type, pollution and biter activity
Pass: optimisations correctly identified; remaining cost correctly
      attributed; a UPS problem is distinguished from a ratio problem

A production-ratio problem that looks like lag, and a UPS problem that looks like a bottleneck, are the two mistakes this course has spent twelve weeks teaching you to tell apart.