Nobody cracks open a programming book anymore

Unquestionably, many people today rely on LLMs more than actually reading. Before that, many depended on the IDE and the web to teach them the language. (What taught them the paradigm?). I think one of the issues though goes back to the bookstores themselves. For years, the quality of the STEM books that were actually available in the store declined precipitously. Perhaps it was the Amazon effect. Barnes and Noble and most independent shops didn't understand the market for the professional software developer. Browsing the shelves for "Designing Data-Intensive Applications" or "Clean Code" and instead finding shelves of Dummies books and IPhone for Seniors, etc. drove buyers to the web and Amazon. Even now, B&N Online can't match Amazon for selection. Since demise of the Borders chain, bookstores have been unwilling to absorb the carrying costs for deeper books that sell less frequently. The feedback loop kicked in and now the shelves are bare. The one thing that still works for me at the local B&N is watching for special orders that were declined by the customer and subsequently shelved. That's how I scored the latest version of the Art of Electronics for 40 something dollars. Those of you who are younger and rely on the web and "AI" will someday realize that your knowledge is less structured, and probably more shallow, than those of us here who are dinosaurs and learned the old way with textbooks and the like.

Here's a brief summary from Sears' archives.[1] I read a much better description in a business book once, but it's not online.

Here's the concept. The simple way to fulfill orders is Doordash Shopping. One person goes to a store and picks the entire order. Performance degrades with the size of the store. This takes O(orders * inventory size) effort. If too many shoppers are trying to fill orders, they get in each other's way and the building traffic starts to choke. This is why Doordash Shopping has a big markup vs. Amazon.

The classic Sears schedule system takes the incoming order and breaks it down by department and picker. Pick slips are generated, originally by hand, and sent to the various pickers, originally by pneumatic tube. Pick slips have a 45 minute time window and a destination bin number. Pickers get pick slips, all of which are for items in their area, grab the item, and put it on a conveyor with the pick slip attached. The conveyor leads to the destination bin area, where orders are assembled in the bins. At the end of a 45 minute cycle, all bins are sent to order checking and outprocessing, and new empty bins are placed for the next fill cycle.

The order paperwork is already attached to the bin, so the checker checks off the items in the bin vs. the items on the order. Excess items from picker errors become go-backs; missing items are not charged to the customer. The total is computed and billed. The bin and paperwork go off to packing and shipping.

Each picker moves only short distances. No running around the giant warehouse. That gets the overhead down to O(orders * log(inventory size)). Pneumatic tubes carry the data that drives the pickers, and conveyors do most of the carrying. The system can develop a backlog in the incoming order area, where it's obvious and can be managed. The picking areas don't usually choke, because their load is bounded by the number of bins per cycle.

It was all done with paper, forms, carbon paper, clipboards attached to totes, cash registers, conveyors, and pneumatic tubes. Julius Rosenwald (1862-1932) designed the original system and became quite wealthy.

Sears should have become Amazon. They had the fulfillment system. They had catalog ordering working well. They never got the online front end right.

[1] https://www.searsarchives.claeys.co/history/history1900s.htm...

Still, when ß is not available one can use ss to represent it. Same with l and ł etc