Single member district election systems are inherently biased outside certain rare distributions of voters that usually aren't present.
Maximal bias can be prevented with historic voting blind formulas for drawing districts, but minimal bias consistent with single member districts (i.e. gerrymandery bias free results) can't be achieved without considering historic voting practices.
Single member district election systems trade systemic regime stability enhancements that it provides, for a less accurate reflection of the public will than proportional representation systems, and whether that is worth it is a value choice.
The Inherent Biases Of Single Member District Only Systems
A system of exclusively electing legislators to a parliament or multi-member legislature from single member districts of approximately equal population is inherently biased.
It is biased against homogeneously dispersed minority factions, even if they are substantial. It is biased in favor of factions that have majorities in geographic concentrated areas (and especially in favor of factions that have majorities in geographic concentrated areas that co-exist with minorities in another faction or factions in the same area).
This bias relative to proportional representation is potentially present in almost all case except those where almost everyone in any given location favors just one dominant political party and the population of those pockets of support for a political party are quite large relative to the population of a typical legislative seat.
When this condition does not hold, a match between a pure proportional representation outcome and the actual allocation of elected officials by party is extremely difficult to secure unless that regions where this doesn't hold almost exactly balance each other out and you have a two party system.
A mathematical formula for drawing boundaries is generally insufficient to prevent this bias from emerging.
There exist maps that minimize the bias that arises from single member district systems relative to proportional representation systems that have only the bias completely inherent in a single member district system. Arguably, when you talk about a map not being gerrymandered, in a context in which a single member district system is a foundational assumption, this is what one means.
But, it is not possible, in general, to minimize that bias merely from knowing the geographic distribution of people on a map. Without knowing their historical partisan preferences, no formula consistent or almost always minimizes the map's bias relative to a proportional representation system.
You can use a "voting history blind" formula to prevent a maximal bias relative to a proportional representation benchmark, but you can't minimize it.
Drawing perfect districts is even harder when accurately representing the relative power of the political parties competing is not the only goal.
For example, the districts that maximize that goal are not the same as the districts that maximize ethnic diversity in the legislative body for which the elections are held are both are legitimate considerations.
These conflicts between competing goals are not nearly so stark in proportional representation systems.
The Case For And Against Single Member Districts Only
There are still arguments in favor of single member districts.
Not all factions present an equal threat to the stability of a state. A faction holding a majority support in a geographically contiguous area is a secession and insurgency threat, even if that localized majority is a small share of the nation's total population. So, it is important that such regions perceive that they are well represented in the overall national political process. In contrast, a faction that is a much larger share of the nation's total population, but is a minority everywhere rarely presents a secession and insurgency threat, so it is less important from a national stability perspective to give that dispersed large majority a full political voice relative to its numbers.
A single member district, plurality vote system is also very simple to understand and administer. You count votes in each district which is independent of every other district. The person who gets the most votes wins.
Furthermore, in the vast majority of those districts, the outcome won't be remotely close. You have close votes that change control of the country and really matter only when the competing legislative coalitions are very close to 50-50 (which admittedly a two party system naturally gravitates to over time) and in which the swing districts are very close to 50-50. But, in if that happens, a disputed election boils down to just a very simple counting process in just a handful of close races, at a time when the country is almost equally divided between two major parties or coalitions. Both the confined nature of the bona fide disputes and the simplicity strongly disfavor outcomes where a credible election contest is possible, and particularly at fragile moments for a nation's survival, clarity of succession can be more valuable than accuracy (especially when almost exactly half of the country favors each side).
Closely related to this point is that when the nation is not almost equally divided 50-50 and one party instead has an exaggerated edge, the winning party will tend to have a legislative majority that is much safer than their electoral majority. This system over rewards winners and over punishes losers. This, in turn, makes it easier for the winning party or coalition to govern after the election in a stable way. Razor thing legislative majorities like the one that the U.S. has now in Congress are rare.
Also related to that point is that single member district systems strongly favor the development of a two party political system. Two party political systems are much more prone to having clean majorities for one party or another after an election than systems with three or more parties. A single member district system forces politicians to form their coalitions before the election rather than after it to get elected. So, post-election delays in determining which party is in control that have been common in Belgium, Israel, and historically, in Italy, and which are currently an issue in Germany, rarely arise, avoiding another form of potential instability and uncertainty.
And, between elections, a single member district system provides a very direct and clear avenue for a citizen to complain about the government from someone who is more likely than not to be sympathetic to them, without regard to who is currently in power. This sense of being heard, by a particular person who is responsible for them also can reduce the sense of futility that can lead to insurrection and government instability.
In a single member district system you are essentially gaining a system that favors the stability of the regime in the short run, over a system that more accurately reflects the wishes of the population as a whole. Whether the tradeoff is worth it or not is ultimately a judgment call that doesn't have a right or wrong answer. In theory, modern technology and civil order reduce the risks of instability that single member district systems minimize to a tolerable level.
But the fact that the U.S. has experienced widely believed factually false election disputes even in 2020, and the fact that geographically compact majorities tried to leave the Union in 1861 along geographic lines that still largely match modern political division in the u.S., both suggest that the concerns about instability from a system that isn't as simple can't be lightly disregarded.
Persistent, long term bias of the system towards one party and against the other, which most single member district systems are inherently prone to give rise to, can also eat away at public support for the political system and cynicism in the long run, especially if the ends sought in politics are zero sum and high stakes (like control of the U.S. Supreme Court in the U.S. political system).