{"resolved":true,"match_type":"uuid","source_type":"original_book","book_id":"24410bf1-91f1-4e6b-9613-e64f0132d341","title":"The Stack On Software and Sovereignty","author":null,"edition":null,"publisher":null,"year":null,"isbn":"9780262029575","requested_by":"printed_page","book":{"source_type":"original_book","book_id":"24410bf1-91f1-4e6b-9613-e64f0132d341","id":"24410bf1-91f1-4e6b-9613-e64f0132d341","uuid":"24410bf1-91f1-4e6b-9613-e64f0132d341","title":"The Stack On Software and Sovereignty","normalized_title":"thestackonsoftwareandsovereignty","author":null,"edition":null,"publisher":null,"year":null,"isbn":"9780262029575","isbns":["9780262029575"],"parsed":true,"ai_readable":true,"document_type":"book","page_count":517,"navigation_url":"https://workshop.closeclaw.top/ai/book/24410bf1-91f1-4e6b-9613-e64f0132d341","title_url":"https://workshop.closeclaw.top/ai/title/The%20Stack%20On%20Software%20and%20Sovereignty","fixed_page_url_template":"https://workshop.closeclaw.top/ai/book/24410bf1-91f1-4e6b-9613-e64f0132d341/page/{printed_page}","path_api_url_template":"https://workshop.closeclaw.top/api/ai/book/24410bf1-91f1-4e6b-9613-e64f0132d341/page/{printed_page}","read_url":"https://workshop.closeclaw.top/api/ai/read/24410bf1-91f1-4e6b-9613-e64f0132d341","page_url":"https://workshop.closeclaw.top/api/ai/books/24410bf1-91f1-4e6b-9613-e64f0132d341/page","search_url":"https://workshop.closeclaw.top/api/ai/books/24410bf1-91f1-4e6b-9613-e64f0132d341/search"},"page":{"printed_page":341,"pdf_page":355,"physical_index":355,"physical_index_base":0,"printed_page_status":"recognized","page_sha256":"f46437e5e29d74d5043a5db88b110805e2c570e241af5312cc4045dc708de6e0"},"chapter":"Fheck The Stack |On Software and Sovereignty Benjamin H. Bratton","content":"As we outline the long course of platform governance (both with a nomos of the Cloud and without one) the seat of that governance may be neither in the formal sovereignty of states nor in the headless tournament of markets, but in interfacial couplings themselves. Platforms govern in and as algorithmic decision-making machines. They decide what an interface will or won ’ t do for which User and when. For Agamben, the sovereign is he who decides on the state of exception and on the internal exception of the camp (for him, the “ nomos of the modern ” ) but in the more mundane but pervasive decisions made several times a second by the computationally active interfaces of the User’ s world, the decision over the exception and the reversibility of the interior and exterior is automated. This is the essence of algorithmic governance. In a way, all machines represent ways that we have learned to automate decision and action (and it is at this juncture that the edge of “ the political ” terminates and something else takes over). Laws are also a way to automate decision; they codify decision into precedent, norm, and exact interpretation to be followed over and over again according to code. Laws refer to how interfaces should and should not operate in the future by attempting to articulate in advance all possible variations according to plan, whereas algorithms may automate a decision at the exact point where it needs to be made and where it will affect a User and a User will affect it at the interfacial seam.\n\nIn Stanislaw Lem ’ s Summa Technologica (1964), his treatise on the future of technology (including space travel and things that closely resemble nanotechnology and virtual reality), the Polish science fiction author warns — with clear fascination — against possible totalitarian perils of cybernetic governance ( “ Electrocracy ” ) and what he calls “ The Black Box: The Regulator of the Highest Kind. ” In any conflation of legal regulation, such as a formal statute, with mechanic regulation, such as a thermostat, there always exists the potential for enforcement to become either dangerously opaque, unavailable to redress, or dangerously stupid, demanding compliance even against the logical goals of the regulation, or both. In some recent discussions of algorithmic governance, it has been suggested that such platforms may work best when desired outcomes are well known and understood and when system performance can be quickly measured and finely tuned. <sup>73</sup> Instead of dictating every step along the way, stifling improvisation in the face of contingency, the system is free to find its own best means toward that particular and consensual end. There are, however, at least two problems with this heuristic. The first is if insufficient care is given to who or what is desiring a particular outcome and how these may be determined in general before they are decided in the particular by an interfacial regime. <sup>74</sup> The second is that platforms often just do not work this way. As discussed in the first chapter, a “ plan ” fixes outcomes and may also fix means, a “ bureaucracy ” fixes outcomes and means, whereas a platform can fix means (such as protocols, APIs, data standards) but not the content of outcomes ( “ Google does not ‘ make ’ what you search for ” ). It sets criteria for outcomes, but not the outcomes themselves; that is the point of them. The top search result is the one that algorithms deem the most “ relevant, ” but it could be anything at all (to the chagrin of large content providers that would wish to foreground their highly leveraged content assets). That is, the deciding of outcomes is something that happens in the course of interaction itself and could be a result of the User’ s disciplining the algorithm of automation as much as it is the algorithm delimiting User behavior by what it shows, hides, allows, or prohibits. The relative symmetry or asymmetry of this interfacial relationship is, of course, situation dependent and is itself decisive. <sup>75</sup>\n\nThe Stack to Come\n\n341","previous_page":340,"next_page":342}