guiltygyoza Blog About

Degree of Antifragility // 反脆弱性

Summary: Let’s be more antifragile to innovate faster.

1- As Topology pushes the frontier and explores the unknowns, stressors and how we adapt to stressors seems to strictly determines how fast we learn and grow. Taleb’s teaching feels visceral to us.

2- Let’s coin a term: Degree of Antifragility (DOA). Increasing DOA allows for faster growth. Decreasing DOA leads to longer stasis, staying stuck, being less attentive to lessons and adaptive to feedback, and at worst pressures one to deceive externally and internally, which all lead to slower growth or slow death.

3- One factor that lowers DOA is system coupling. If iterating on A requires changing B, the iteration speed for A is lower than otherwise. (note: in creative industry, low iteration speed is fatal. Topology’s Isaac experiment has high degree of system coupling, which makes iterating on it difficult; in response to the Isaac lesson, Topology’s MovyMovy experiment has low degree of system coupling, allowing for fast iteration on both game mechanics and implementation approaches)

4- Yet another factor that lowers DOA is the pressure to fulfill premature public promises (PPP). Depending on the nature of projects or products, failure to meet PPPs may lead to brand losing trust. The misled desire to not disappoint could lead one organization to making even more PPPs to appease the public - “we won’t deliver X on D day because we decided to take on X^2, a grander project, to be delivered on D+D day”, making it a bigger mess to deal with eventually. (note: projects that have launched tokens by making public promises will have lower DOA - thanks for milan’s comment)

5- Yet another factor that lowers DOA is sunk cost fallacy. The key here is to identify whether a cost is sunk cost that should be severed, or opportunity cost that should be dealt with persistence. Failure to recognize sunk cost fallacy could lead to making PPPs which further lowers DOA and entrenches the organization in the wrong direction; worse, it could lead to internal delusion. If internal delusion is persisted by the leadership at power positions, it could shut down honest voices in the organization, sever its neural system and kill it slowly as resource and morale runs out.

6- Let’s reduce or avoid PPPs. Let’s be mindful of and overcome sunk cost fallacy. Let’s strive to iterate on smaller components in modular systems. Let’s increase our DOA to innovate faster.

(note: other factors that impact DOA include the capital structure of the organization, burn rate, investor expectations, and the ego of its leadership)


1- 隨著 Topology 進行各式鏈上開發實驗,我們發現壓力 (stressor) 以及我們如何應對壓力似乎完全決定我們的成長迭代速度。Nassim Taleb 的教誨如雷轟頂。

2- 讓我們定義一個測度:反脆弱性 (DOA; Degree of Antifragility)。增強反脆弱性帶來更快速的成長。降低反脆弱性使組織停滯期延長、更容易受困、更無法傾聽壓力挾帶的教訓(導致就算換了題目還是會撞到相同的壓力,只不過換了面貌)、無法面對內部和外界反饋,最糟時把組織推向欺瞞 - 向外界欺瞞,甚至向內部欺瞞 - 一切都導致緩慢成長或是慢性死亡。

3- 一個降低 DOA 的因子是系統耦合度。如果為了迭代 A 部件必須同時改變 B 部件,那 A 部件的迭代速度將降低。(註:在講求迭代的創意工作裡,低迭代速度是致命的;Topology 的 Isaac 實驗具高系統耦合度,讓迭代速度慢;MovyMovy 則具備低系統耦合度,因此能夠快速迭代遊戲機制和實作架構)

4- 另一個將低 DOA 的因子是組織被迫實現不成熟的公開承諾 (PPP; premature public promises)。未實現的公開承諾可能帶來品牌失信,至於失信程度要看項目文化或產品本質。為了不失信,組織可能走向錯誤的道路而做出更多的不成熟公開承諾,手腦分離:「我們無法在 D 日交出項目進度 X,是因為我們決定追逐更宏大的進度 X^2,預計在 D+D 日完成。」屆時殘局更難收拾。(註:若項目發了幣,發布公開承諾,將會降低 DOA。感謝 milan 補充。)

5- 還有一個降低 DOA 的因子是沈沒成本謬誤。這裡的關鍵是區分沈沒成本與機會成本 - 是要儘早認賠檢討轉向的沈沒成本,還是要繼續堅持推進所必須肩負的機會成本。陷入沈沒成本謬誤的組織可能會被迫發布不成熟的公開承諾 (註:明明是錯誤的方向卻沒有應變,而繼續向外界承諾能夠撐過去),讓組織陷入更深的泥淖、在錯誤方向上走更遠;更糟的是可能帶來組織內部的妄想現象。倘若妄想發自權力結構中的上層成員,將會熄滅組織內誠實的聲音、切斷組織的神經回饋系統,隨著資源和士氣消耗慢慢殺死組織。

6- 讓我們減少不成熟的公開承諾、更留心和克服沈沒成本謬誤、致力設計有利於高速迭代的低耦合系統。讓我們增加自身反脆弱性以更好的創新。

註:還有其他影響反脆弱性的因子,例如組織資本結構,現金燃燒率,投資人預期,以及領導層的自我 (ego)。