空間垃圾桶模式外在效度之實證檢驗：以台北市為例

Empirical Validation of the Spatial Garbage Can Model: A Case of Taipei

空間垃圾桶模型（spatial garbage-can model）視都市動態系統為一組在空間中隨機游走互動之元素之間交互作用之下的結果，包括決策者（decision makers）、選擇機會（choice opportunities）、問題（problems）、解決方案（solutions）以及區位（locations）。其中決策者、問題及解決方案以不可預測的方式相互碰撞，丟入在特定區位的選擇機會或垃圾桶中而產生模擬都市發展過程的結果（Lai, 2006）。本文嘗試從地區個案資料的分析來檢驗空間垃圾桶模型。具體而言，採用多重個案研究的複現策略，蒐集西門町、信義計畫區與新北投等三個地方的相關資料，透過ANOVA統計檢定分析，探討四種要素結構對系統能量值的影響關係，試圖驗證之前的電腦模擬研究（Lai, 2006）的結果，並進一步計算結構的熵值以觀察這三個地方的都市活動是否具有自我組織（self-organization）的系統特性，且佐以網路科學方法，探討這些都市活動的空間分佈特性。分析結果發現與電腦模擬文獻的結果大致符合：（一）結構限制對系統能量的變化並無顯著的影響；（二）與空間結構相較下決策結構反而顯現出較佳的自我組織特性；以及（三）都市活動網路並未呈現出明顯的小世界（small-world）特徵。此外，研究結果並不違反制度設計對都市發展之影響而言，較空間設計來的更有效的最主要結論。

A dynamic urban system in Spatial Garbage-Can Model (SGCM) is viewed as the outcome of the interaction between a set of elements randomly walking in space. The elements include Decision Makers, Choice Opportunities, Problems, Solutions and Locations. Decision makers, problems, and solutions collide with each other by some unpredictable manner and are thrown into the choice opportunities, or garbage cans, in the particular locations as the main mechanism of the simulation of urban development process (Lai, 2006). This paper attempts to validate externally the SGCM by analyzing three cases in Taipei. In particular, based on the reproduction strategy of multiple case studies through collecting the activity data from Ximending, Xin-Yi Planning District and New Beitou, we conducted the ANOVA analyses to explore into the effects of four structures between the elements on the energy level of the system to verify the results of earlier computer simulations (Lai, 2006). We further calculated the entropies of the structures to observe whether the urban activities in these three places were characterized by self-organization. In addition, we explored into the spatial distribution of these urban activities based on network science. The results of our study were found consistent with the computer simulations: (1) the structural constraints did not have significant effects on the energy level of the system; (2) decision structures tended toward self-organization more significantly when compared with spatial structures; and (3) urban activity network could not be characterized by small-world structures. However, the study did not violate the main observation that institutional design is more effective than spatial design in affecting the urban development process.