Over the past few years, due to the need to migrate legacy systems in enterprise projects, I’ve actually developed little affection for the long tail effect. A project that should have taken only a few months ended up requiring years to complete due to user dependencies or stubborn system coupling.
However, my perspective on this has shifted recently due to AI’s rapid transformation. The long tail may not necessarily be a bad thing. The main reason is that I’ve come to realize: as AI gradually replaces human programming, people are losing technical skills and can no longer code with the familiarity they once had. Yet in reality, most people still live in what we might call the old era, manually and inefficiently completing various tasks. From this perspective, AI has become the new infrastructure. If some force causes this infrastructure to collapse, then those who have walked at the forefront of the era and thus lost their skills would paradoxically need to seek help from those still living in the old era to solve coding problems.
If those walking at the head of the adaptation curve ultimately fail to survive the long course of human history, then those who haven’t yet adapted to the future might instead forge an alternative path toward a different future.
From this perspective, I now better understand why in many science fiction works (such as Cloud Atlas), future humanity possesses advanced technology yet lives extremely primitive lives.
The following content is generated by LLMs and may contain inaccuracies.
Context
This idea touches on the “skill fragility paradox” in technological evolution—when a technology becomes infrastructure, the populations depending on it may become more vulnerable in the face of systemic risk. In the current rapid penetration of AI into programming, this problem extends from the engineering dilemma of legacy system migration to the evolutionary strategy of human skill composition. It echoes Nassim Taleb’s core argument about antifragility: excessive optimization weakens the ability to adapt to sudden change.
Key Insights
-
Skill Redundancy as Survival Insurance — The phenomenon you’ve observed validates the “Collingridge dilemma”: technology is easy to modify early but its impacts are difficult to measure; once mature, impacts are clear but the technology becomes hard to change. Venkatesh Rao’s analysis on Ribbonfarm points out that “laggards” on the technology adoption curve actually maintain diversified skill repositories, which in black swan events can transform into critical advantages—analogous to how biodiversity contributes to ecosystem resilience.
-
Hidden Dependencies of Infrastructure — AI-enabled programming is creating new forms of “technical debt.” When coding ability is outsourced to AI, we face not merely skill atrophy but the comprehension gaps created by cognitive offloading. Historical cases like GPS causing spatial cognition decline, or calculators affecting mental arithmetic abilities, demonstrate that convenience tools reshape rather than merely enhance human capabilities.
-
Prophetic Insights in Science Fiction — The “Cloud Atlas”-style “high-tech primitivization” is not a paradox but an artistic expression of the risks of over-specialization. Joseph Tainter argues in The Collapse of Complex Societies that the maintenance costs of complex systems may ultimately exceed their marginal benefits, leading to simplification and reversion. Your legacy system migration predicament is, at the macro level, a microcosmic manifestation of this complexity trap.
Open Questions
- As AI tools proliferate, which “inefficient” manual skills are worth deliberately preserving as strategic redundancy? How should we quantify the value of such insurance?
- If we view the human skill ecosystem as an investment portfolio, how should the optimal “long-tail/head” allocation ratio dynamically adjust with the speed of technological change?
最近几年，因为企业项目需要迁移遗留系统，我对长尾效应反而没有太多好感。本来几个月就能完成的项目，却因为用户依赖或顽固的系统耦合，导致整个迁移需要数年才能完成。
不过，最近因为AI的快速变革，我对这件事的认识有所转变。长尾可能并不一定是坏事。主要原因是我意识到：AI逐步替代了人类编程，导致人们在技能上的丧失，不再能像从前那样熟悉地编码。但实际上大部分人还生活在所谓的旧时代，手动低效地完成各种任务。从这个角度看，AI成为了新的基础设施。如果某种力量导致这种基础崩塌，那么这些走在时代前沿因而丧失了技能的人，反而会因为缺乏解决编码问题的能力，转向寻求那些还留在旧时代的人的帮助。
如果这些走在适应曲线头部的人最终没能在人类历史的长河里生存下来，那些还没来得及适应未来的人反而另辟蹊径，走向其他的未来。
从这个意义上，我更能理解为什么在许多科幻作品（比如《云图》）中，未来人类掌握了许多尖端科技，反而过着极其原始的生活。
以下内容由 LLM 生成，可能包含不准确之处。
Context
这个想法触及技术演化中的"技能脆弱性悖论"——当某项技术成为基础设施后,依赖它的群体反而可能在系统性风险面前更加脆弱。在AI快速渗透编程领域的当下,这个问题从遗留系统迁移的工程困境延伸到了人类技能组合的演化策略。它呼应了Nassim Taleb关于反脆弱性的核心论点:过度优化会削弱适应突变的能力。
Key Insights
-
技能冗余作为生存保险 — 你观察到的现象印证了"Collingridge困境":技术早期易改但影响难测,成熟后影响明确却难以改变。Venkatesh Rao在Ribbonfarm的分析指出,技术采纳曲线上的"落后者"实际维持着多样化的技能库,这在黑天鹅事件中可能转化为关键优势——类似生物多样性对生态系统韧性的作用。
-
基础设施的隐性依赖 — AI编程正在创造新的"技术债务"形式。当编码能力外包给AI,我们面临的不仅是技能退化,更是认知卸载带来的理解断层。历史案例如GPS导致的空间认知能力下降,或计算器对心算能力的影响,都显示便利性工具会重塑而非仅仅增强人类能力。
-
科幻中的预言性洞察 — 《云图》式的"高科技原始化"并非矛盾,而是对过度专业化风险的艺术表达。Joseph Tainter在《复杂社会的崩溃》中论证,复杂系统的维护成本最终可能超过其边际收益,导致简化回归。你的遗留系统迁移困境,在宏观层面就是这种复杂性陷阱的微观映射。
Open Questions
- 在AI工具普及的过程中,哪些"非效率"的手工技能值得刻意保留作为战略性冗余?如何量化这种保险的价值?
- 如果把人类技能生态看作投资组合,最优的"长尾/头部"配置比例应该如何随技术变革速度动态调整?