Back to home

Compare

Comparing: Dwarf Fortress Caves on Git After 20 Years — A Hardcore Dev's Compromise & Dwarf Fortress 用了 20 年才接受版本控制 — 硬核开发者的妥协

AEN
Dwarf FortressTarn AdamsSimon Willison·

Dwarf Fortress Caves on Git After 20 Years — A Hardcore Dev's Compromise

We've been tracking this story: Dwarf Fortress, in development for 20 years and possibly the most complex codebase in indie gaming, only accepted version control (i.e., code management tools like Git) in 2023. His reason was deeply personal: "I don't like my code being thrown into a black box."

What this is

Dwarf Fortress is a deep simulation game that Tarn Adams has led solo since 2006, notorious for its extreme complexity — each world consists of hundreds of layers of historical simulation, thousands of inhabitants, and tens of thousands of items, with nested AI behavior trees. Tarn long refused version control because, in his words, he "didn't like code being thrown into a black box without an immediate sense of return."

Only in 2023, just months before the Steam release, did he admit in an interview: "Now I have to scroll through Discord channels, I have to use version control — the workflow has genuinely gotten more complex." A 17-year-old indie title with staggering code volume finally bowed to modern engineering practice.

Industry view

The mainstream reaction skews more admiring than critical. Simon Willison wrote on his blog: "I'm a bit sad they didn't hold out until they were millionaires before caving, but the persistence itself is admirable."

But there is pushback. Some engineering-minded readers point out that 20 years without version control means bug-fix backtracking is nearly impossible — the project ran more on the miracle of single-developer memory than on engineering capability. In other words, this is individual heroism, not a replicable methodology. Apply the same workflow to any collaborative team of more than five people, and it collapses outright.

Impact on regular people

For enterprise IT: A reminder that tool selection is about more than efficiency — it's about friction between personal preference and organizational culture. Forcing new tools doesn't always land; giving key people transition runway is more realistic.

For individual careers: Sticking to a contrarian workflow for 20 years is itself a rare signal — but even rarer is knowing when to abandon it.

For consumer markets: Players paying over a million dollars for the Steam version proves "geek aesthetics" have real commercial space — you don't need to abandon hardcore depth to chase the mainstream.

BZH
Dwarf FortressTarn AdamsSimon Willison·

Dwarf Fortress 用了 20 年才接受版本控制 — 硬核开发者的妥协

我们注意到一件事:开发了 20 年、代码复杂度可能是独立游戏之最的 Dwarf Fortress,开发者 Tarn Adams 直到 2023 年才接受版本控制(也就是 Git 这类代码管理工具)。他给出的理由很私人:"不喜欢代码被丢进一个黑盒子里"。

这是什么

Dwarf Fortress 是 2006 年起 Tarn Adams 一人主导开发的深度模拟游戏,以极端复杂著称——一个世界由几百层历史模拟、几千名居民、上万件物品构成,AI 行为树层层嵌套。Tarn 长期坚持不用版本控制,因为"不喜欢代码被丢进一个黑盒子里,没有即时回报感"。

直到 2023 年 Steam 发售前几个月,他在访谈中承认:"现在要查 Discord 频道、要用版本控制了,流程确实变复杂了。"一款上线 17 年、代码量惊人的独立作品,最终还是向现代工程实践低了头。

行业怎么看

主流反应是敬佩多于批评。Simon Willison 在博客里写道:"有点难过他们没撑到百万富翁才让步,但这个坚持过程令人佩服。"

但也有反对声音。一些工程派读者指出,20 年没有版本控制意味着 bug 修复回溯几乎不可能,项目能跑下来更多依赖单人记忆的奇迹,而非工程能力的体现。换句话说——这是个人英雄主义,不是可以复制的方法论。放到一个 5 人以上协作团队,这套玩法直接崩盘。

对普通人的影响

对企业 IT:再次提醒我们,工具选择不只关乎效率,也关乎个人偏好与组织文化的摩擦——强推新工具未必能落地,给关键人物留过渡期更现实。

对个人职场:坚持一种"反共识"工作方式能撑 20 年,本身是一种稀缺信号;但更稀缺的是认清何时该放弃它。

对消费市场:玩家为 Steam 版付费超百万美元,证明"极客审美"也有真实商业空间——不必为了大众化放弃硬核。