SAME UNIVERSITY CITY · 81 YEARS · THREE ROUTES · THREE READINGS
Cambridge, Massachusetts — Logical Switches, Public Keys, and Auditing Faces
Cambridge, MA·1937 CE → 2018 CE·3 routes · 3 readings
In 1937 Shannon translated between Boolean expressions and relays; in 1978 RSA published an asymmetry between public and private exponents. In 2018 Gender Shades split overall face-analysis accuracy into intersectional errors, widening the question from mathematics that builds systems to mathematics that audits them.
QUESTION FOR THIS CROSSING
How should the power to design circuits and keys with symbols be joined to the responsibility to compare where a system fails across faces?
COMPARISON BOUNDARY
One university city across eighty-one years proves neither a continuous team nor inevitable progress. Each card retains independent switching work, public-key precedents, RSA implementation conditions, and the dataset and commercial-classifier scope of Gender Shades.
ONE CROSSING WITHIN 21
Keep the surrounding geography visible
Every marker stays visible. The amber marker is the current place; the dotted line remains a viewer itinerary rather than a historical route.
지도를 불러오는 중…
Same place · different time layers
Compare the route readings
Shannon asks about correspondence between expressions and relay states; RSA about asymmetric modular operations; Gender Shades about benchmark construction and intersectional performance gaps. Design and audit are not made one algorithm or direct transmission.
1. When Symbols Became Machines — From Written Procedures to Stored Programs
1937 CE · Composition
Shannon — Turning Logical Expressions into Relay Circuits
MIT graduate student Claude Shannon connected open and closed relays with Boolean true and false, allowing complex switching circuits to be analysed and simplified as logical expressions. His 1937 thesis is distinct from its 1938 publication, and Victor Shestakov and others produced independent related work. The achievement was a design bridge, not a literal identity of logic and electricity.
QUESTION FROM THIS ROUTE
Can a circuit of hundreds of relays be analysed and redesigned more simply through logical expressions instead of tracing every wire?
What became newly visible
Shannon mapped open and closed relay contacts to Boolean zero and one conditions, writing circuits as expressions and simplifying them by equivalence transformations. A two-way design translation between symbolic logic and physical switches turned circuits from trial-and-error wiring into calculable structures.
Why this place could enable it
MIT electrical engineering and maintenance work on Vannevar Bush's Differential Analyzer gave Shannon a concrete problem in complex relay circuits, while academic and industrial networks circulated the 1938 paper. Shestakov and others produced independent related work elsewhere in the same period.
What actually moved
Symbolic logic after Boole → relay-circuit problems in MIT's Differential Analyzer → 1937 master's thesis and 1938 AIEE paper → switching and communication research at Bell Labs → digital logic synthesis and hardware design
DO NOT OVERREAD THIS SCENE
This was not discovery of a natural identity between logic and electricity but a model of specified relay conditions in Boolean expressions. The 1937 thesis is distinguished from its 1938 publication, and independent prior or parallel work by Shestakov and others is not erased.
2. Cities That Kept and Broke Secrets — From Letter Frequencies to Public Keys
1978 CE · Composition
Easy to Multiply, Hard to Reverse — RSA Public Keys and Signatures
At MIT, Ronald Rivest, Adi Shamir, and Leonard Adleman used a composite made from two large primes and modular exponentiation to give a concrete method in which a public exponent encrypts or verifies while a secret exponent decrypts or signs. A key for everyone and a key held by its owner became distinct. Textbook “plain RSA” is not safe for direct modern use: padding, authentication, key generation, and the surrounding protocol all matter.
QUESTION FROM THIS ROUTE
Can one pair of computational keys let anyone lock a message for its owner and anyone verify a mark only the owner could make?
What became newly visible
The asymmetry between easily multiplying two large primes and hard factorization, together with modular inverses, creates public and private exponents. Encryption and signing become opposite directions in one algebraic structure, giving a concrete way to distribute keys through a public directory.
Why this place could enable it
MIT’s computer-science community rapidly made the problem opened by Diffie–Hellman concrete through three researchers’ different intuitions and public review. Cambridge university, journal, and software cultures let the method be reproduced and attacked outside secret agencies.
What actually moved
Diffie–Hellman’s public-key and signature agenda → MIT collaboration by Rivest, Shamir, and Adleman → combination of Euler/Fermat results with primes and factorization → 1977 report and 1978 CACM paper → standards, libraries, certificate ecosystems, and later attacks
DO NOT OVERREAD THIS SCENE
RSA security is not automatically proved by the sentence “factoring is hard”; key size, randomness, padding, implementation, and use matter. Deterministic plain RSA is not used directly in modern practice, and internet security is not reduced to one RSA invention.
3. When Data Began to Read People — Who Was Counted, Who Disappeared?
2018 CE · Performance audit
Recounting the Faces Hidden under Average Accuracy — Gender Shades
Joy Buolamwini and Timnit Gebru evaluated commercial gender classifiers separately across intersections of perceived skin type and gender. Large performance gaps and the lighter-skinned, male-heavy composition of existing benchmarks became visible beneath one average accuracy. The study audited four classifiers and constructed datasets at a particular time; it was neither a theory of every face-recognition system nor a justification for reducing gender identity to a binary classification task.
QUESTION FROM THIS ROUTE
Even when overall accuracy looks high, how can we reveal which intersectional group bears most errors?
What became newly visible
Move from one average score to error tables disaggregated jointly by perceived skin type and gender, making dataset composition and performance gaps auditable.
Why this place could enable it
MIT Media Lab's research and public-paper setting, plus a new benchmark built from parliamentarian images, enabled external comparison of commercial APIs.
What actually moved
Audit existing face datasets → Pilot Parliaments Benchmark → intersectional evaluation of commercial classifiers → public paper and company response → algorithm-auditing movement
DO NOT OVERREAD THIS SCENE
Results for particular 2018 gender classifiers are not permanent rankings of all face recognition, nor do they naturalize binary gender classification itself.