In 2011, I learned, via internet, about Toshers.
https://retroworks.blogspot.com/2011/11/history-epic-everyday-waste.html
I wrote about an uncontrollable new resource - water creating flush toilets - creating incredible convenience and applauded demand - overwhelming the waste management systems.
Unpopular buggy-whip-opinion hypothesis: Most of the information we observe is actually a waste product of existing stuff.
"Flush toilets created volume that overwhelmed dung pits and toshers in London. AI is going to create an "information volume" (true and false info) which will deluge human capacity to digest it, same as toshers were put out of livelihood by flush toilets."
In 19th-century London, the introduction of the water closet did not merely improve hygiene; it multiplied the volume and liquidity of human waste far beyond what cesspits, night-soil men, and the informal economy of toshers could absorb. The old system was labor-intensive, localized, and semi-closed. The new one was high-volume, continuous, and forced a total redesign of infrastructure (Bazalgette’s sewers) while rendering the previous scavengers economically and practically obsolete. Toshers did not disappear because someone banned them; they disappeared because the medium itself changed.
AI is doing something structurally similar to the information environment.
Information is reused, recycled, observation. Its quality is similar to the recycling stream of solid waste. What do people sort correctly, and what information is identified as valuable?
It does not just accelerate the production of true statements. It generates a continuous, low-cost flood of text, images, audio, and video—much of it plausible, some of it correct, a large fraction of it synthetic or adversarial. The human cognitive and institutional apparatus that previously filtered, verified, summarized, and valued information (journalists, researchers, editors, librarians, teachers, even attentive citizens) was calibrated for a lower-volume, higher-friction regime. That apparatus is now being asked to process an order-of-magnitude increase in throughput while the signal-to-noise ratio deteriorates. The result is not merely overload; it is a change in the medium that devalues many of the old skills and roles.
Several differences matter: Asymmetry of cost. Generating another plausible paragraph or deepfake is now near-zero marginal cost. Verifying it is not. The same asymmetry existed when sewage was freely available to flush.
Feedback loops. False or low-value information does not simply sit there; it trains further models, shapes attention, and alters incentives. Sewage did not train the next generation of sewage.
Persistence of demand for truth. Societies still need accurate maps of reality for coordination, science, law, and survival. The deluge does not eliminate the need; it raises the price of obtaining it. New intermediaries (better search, verification tools, specialized human judgment, reputation systems, cryptographic provenance) will arise, just as modern sanitation engineering arose. Some of those intermediaries will themselves be AI-mediated.
Livelihood displacement. Roles whose primary value was intermediate filtration, summarization, or first-pass synthesis are already under pressure. The analogy to toshers is closest here: the work does not vanish because it is forbidden, but because the volume and form of the material make the old methods uneconomic or irrelevant.
The historical parallel suggests two broad outcomes rather than simple collapse. Either societies invest in new infrastructure capable of handling the volume (technical, institutional, and cultural), or they tolerate a degraded information environment in which most people operate with noisier maps and a smaller set of trusted sources. Both happened, unevenly, with sanitation. The same is likely here.
The core insight is correct: once the volume of the medium itself changes by an order of magnitude, the previous ecology of handlers is not merely stressed; it is structurally displaced.


No comments:
Post a Comment