In most RTS games a unit producing structure is able to train a single unit at a time. This "serial" unit production is how all Terran and Protoss unit producing structures work[1]. Because the next unit can not start construction until the previous unit has finished, minor hiccups in keeping the structure busy are disproportionately punishing to players.
In Starcraft 2 a worker gathers 40 minerals per minute and takes 12 seconds to train. If you leave a worker idle for 12 seconds you lose 8 minerals. If you leave a command center idle for 12 seconds you have one fewer worker than you would have had with constant production, so you lose 40 minerals per minute for the rest of the game. I think this extreme punishment is the single biggest problem with RTS accessibility. It is the reason why a Diamond level player is comfortably able to 1v4 Gold level players. It is the reason why nearly every game from Bronze up to Masters league is decided by probes and pylons. It is the reason why a not insignificant number of pro games end with one player anticlimactically steamrolling the other player in a single push despite the defender making all of the correct strategic decisions.
Some people have proposed solving this by letting people automake workers. I don't like that solution. I think players should gain advantages for managing their economy better than other players. The problem is that the current advantages are way too extreme. Even if you don't accept that argument, surely the response from the core RTS audience every time automake is proposed will give you pause. It is not the right hill to die on.
Some people have proposed letting people automake workers, but making it less efficient (say 20%-30%) than queueing workers manually. I don't like that solution either because it creates something analogous to a welfare cliff[2]. When an intermediate player is at a skill level where they are ready to make the transition to manual worker queueing their skill level will drop, at least temporarily, as they adjust to the new more difficult way to do things.
A better alternative is to look at how Zerg unit production works. Ignoring inject, the hatchery passively produces larva over time. The player can return to the hatchery whenever they choose to train as many units as they have larva simultaneously. This means that as long as a Zerg player does not hit the larva cap they are not punished as harshly as the other races for idling their production structure. If you do not cap on larva, forgetting to morph a larva into a drone for 12 seconds is the equivalent of idling a worker for 12 seconds. That means its only costs you 8 minerals instead of the 40 per minute it costs the other races[3].
How could this model be applied to other races? Imagine if Terran Command Centers passively generated "colonists" every 12 seconds, and a player could turn all of their pooled colonists into SCVs at once. Unlike larva, Colonists couldn't be used for anything else except SCVs. Now if a player briefly forgets to build SCVs they don't permanently fall behind in income by massive game ending amounts. Instead they pay an opportunity cost for not working that is equivalent to leaving SCVs idle. If the designers don't believe that punishment is harsh enough on its own they now have new avenues for adjusting that cost too. For example, Colonists could charge upkeep while they exist that SCVs do not. Alternatively the spawn rate of colonists could increase geometrically as you pool more and more so that a big lapse in macro is more punishing than a minor one. Either of those punishments would be massively less brutal than the one that currently exists for Terran and Protoss.
I acknowledge that the solution I'm proposing feels kind of "gamey", but I think it has the potential to significantly improve multiplayer accessibility by raising the skill floor. Now when a median player plays a top decile player their army might be only 90% as large as their opponent's instead of 25% as large. Thats still a game-losing disadvantage but its no longer a "Why am I even playing this game" level disadvantage. I believe it would also decrease the frequency of one-sided pro games where one player makes a minor mistake and instantly loses. Best of all, because there is still an advantage to be gained by crisply making workers, it would preserve the macro skill ceiling.
Even if you don't like the idea I'd be curious to hear your thoughts on the serial worker production problem as I've described it.
TLDR: All Town Centers should make workers in parallel the way hatcheries do.
[1] Yes, the SC2 terran reactor technically makes two units at once. I'd argue it is effectively just two barracks glued together. [2] https://en.wikipedia.org/wiki/Welfare_trap [3] I think this is the real reason Blizzard couldn't automate inject larva during LotV. Forgetting to inject is comparably punishing to forgetting to build SCVs. Zerg would be by far the easiest race if instead they were paying handfuls of minerals for forgetting to build workers while other races were paying the massive amounts I've described.