A worker rush is a cheese wherein you take all of your starting SCV's/Probes/Drones to the enemy base and attack. It can be done with any race, though Zerg is typically the strongest, with Protoss in second, both due to their natural regeneration. Terran is typically weakest, as Repair requires minerals, which thus requires at least 1 SCV left behind.
The problem with a worker rush is that if your opponent has built even one more worker, your attack force will be outnumbered. The correct counter in most cases, is for the defending player to select all their workers, attack move, and win via numbers. There does not tend to be much art to microing worker combat, so it's difficult to claim to consistently defeat 8 enemy workers with 6 of your own.
There is exactly one build which this approach might legitimately counter: a 6 pool. As the Zerg player will not build workers, and further drop to 5 workers after placing the pool, a worker rush - especially with a 7th helper worker sent along afterward - might work given a short rush distance. Of course, by definition, 6 pools cannot be scouted, so this relies on knowing your opponent and reading their strategy from the onset. Against any other build, the only hope is a distracted / confused defender who does not set your opponent's workers to attack.
This should not be confused with worker harass, where a scout attacks workers (in particular, SCVs) in the opponent's base to distract them and possibly force a stop on the building.
Alternatively versus Terran on maps with short rush distances like Steppes of War, Zerg can worker rush with 12 Drones by doing a double Extractor Trick. This exploits the timing window where Terran cannot build workers for a few seconds before their Supply Depot finishes as well as denying their Barracks. Since buildings like Gateways and Spawning Pools cannot be denied effectively, this strategy is unlikely to work versus Protoss or Zerg.
All-In to destroy the enemy base immediately despite being outnumbered.
This build is countered by pretty much any of:
- Build workers. Since almost every strategy involves producing at least some economy early, this is easily done. If you see 6-7 workers entering your base, select all of your own, perhaps ball them up, then attack move. If only 6 workers came and they start dancing, go back to mining. You should usually keep producing workers while possible, as well as utilising any production facilities you may have produced before the fight.
- An early wall-in, though this is unlikely to finish unless the rush distance is very long.
- If you were 6-pooling, it's probably worth it to cancel the Spawning Pool and immediately build 3 more Drones with the money from the cancellation. Have your 6 (5 mining + 1 from cancel) Drones run and stall until the additional Drones spawn. The Drones will be more helpful and come faster than any Zerglings, plus, you might not even have enough minerals to build the Zerglings.
Attack with all starting Workers and hope. This all-in relies heavily on lots of surrounds to even the odds quickly or an inexperienced opponent. You should train a 7th worker and rally it to the enemy base. There is little point in attempting to use the 7th to mine; either you wipe out all the enemy workers and win, or your attack force dies and you lose. If the enemy scatters your opponent's workers away from their base, then just stay there and burn it down until they return. If they are Terran and blast off, well, you can move your workers back to your base and go mine there, while making sure your opponent does not land.
A Protoss player who worker rushes can attempt to run away and wait for shields to recharge after doing some damage, only to return when the victim returns to mining, since all workers move at the same speed.
As a Terran, using a single SCV to Repair in a battles with equal worker number will tip the scales in your favour. Using 2 to repair will be detrimental and using repair when fighting with fewer numbers will put you at an ever bigger disadvantage.
If at any time you have significantly more workers than your opponent (for example, 5 to 2), you can attempt to retreat and resume mining with the economic advantage. If you do, prepare yourself for a slow and boring game. Or you could just attack their remaining workers and base until they quit.
- A good map for this strategy would be a map with a short distance between bases. A long rush distance makes this build worthless. Steppes of War would be an example of a map where this all-in may have a better chance of working.
- On maps with multiple starting locations, the attacker guessing the wrong enemy base has almost certainly lost.
This strategy was used to get the "Dark Voice" portrait first (1,000 wins as each race + 1,000 random wins), as these matches last about 3 minutes maximum and Bronze League players did not always realize they could have their workers defend themselves. The (in)famous player first known on the official battlenet forums named Pennoyer achieved the "Dark Voice" portrait using this strategy. Past Bronze League, this strategy really should never work at all, barring the rare "Aha I know you're going to 6 pool on Steppes of War" case.
IdrA versus Allaboutyou