July 23, 2023

Reviewing The Box

In 2022, I read The Box - a book about the invention and standardization of the shipping container.

I think some approaches and organisational principles carry over from the shipping industry into software development.

Below are snippets from the book and my comments.

But after years behind the wheel themselves, the McLeans understood that the easiest way to control costs was to get employees involved. Holding down insurance and repair bills. for example, meant having safety-conscious drivers. Novices were trained by being paired with senior drivers on the run from Winston-Salem to Atlanta. The senior driver got a bonus of one month's pay if a man he had trained made it through his first year without an accident. The incentives were powerful: the veteran had a strong financial incentive to train the newcomer well, and the new driver understood that he had best drive very carefully if he wanted to stick around.

Formal mentoring with an explicitly rewarded set of behaviours, where result can be objectively measured. Curious if any technology companies assess the quality of mentorship by measuring the performance of the mentee on the job.

Malcom McLean's fundamental insight, commonplace today but quite radical in the 1950s, was that the shipping industry's business was moving cargo, not sailing ships. That insight led him to a concept ot containerization quite different from anything that had come before. McLean understood that reducing the cost of shipping goods required not just a metal box but an entirely new way or handling freight.

Your customer cares about moving solving their problems i.e. moving their cargo, not your ability to sail ships. The public interface of a shipping company is "move my cargo as quickly and cheaply as possible from A to B". Once you offer that service to your users, you are free to implement the solution the way you see fit - invent containers, rebuild port infrastructure to load and unload them, change your ships from breakbulk into container shipping. If you become successful some of your implementation detail will inevitably leak out - people made base their decisions based on what they know about your service.

Weldon's research department pursued its quest for optimality by investigating the most efficient way to use Matson's fleet. Renting time on an IBM 704 computer at several hundred dollars a minute, the researchers built a fully fledged simulation model of the business, incorporating data on volume and costs for more than three hundred commodities at every port the company served at every time of year. Then they added in data on port labor costs, the current utilization of docks and cranes, and the load aboard each ship, to provide real-time answers to practical questions:

  • Should a big Hawaii-bound ship call at Hilo and Lanai, or should it transfer its cargo to a feeder ship at Honolulu?
  • What time of day should a vessel depart Honolulu so as to minimize total costs of delivering a load of pineapples to Oakland?

Where there is a will to implement data-driven decision making, there is a way, even if it costs hundreds of dollars a minute. The company committed resources to rent "cloud computers" to get insights that can increase its revenue. Simulated novel situations when they didn’t have data. Similarly to trading firms spending millions to investigate new trading strategies or FAANG companies training neural networks to optimise MAUs/other metrics.

They were just recruiting … Many were invited to Newark without being told what job McLean had in mind for them. When they arrived, they were given intelligence and personality tests - a rare practice in the 1950s. McLean wanted people who were smart, aggressive, and entrepreneurial, the wrong test scores meant no job offer. Education did not matter;

Hired for brains with an equal opportunity test regardless of one's educational background. Assessed culture fit with a personality test. While technical aptitude remains a uniformly administered interview round these days, culture fit is less explicit.

although Malcom McLean had a box at the Metropolitan Opera, intellectual airs were frowned upon, and new hires were advised to fracture their grammar to fit in with a crowd of truckers. "When we had nothing else to do, we would stand and pitch pennies," remenbered naval architect Charles Cushing, an MIT graduate who joined the company in 1960. "They don't teach you to pitch pennies at the Wharton School.»

Know Your Customer and fit in. Employees were explicitly told to "culture fit" with their customers - truckers.

The work was seven days a week, in an exciting, demanding environment. Memos were not wanted. Conflict among executives was a given; managers were expected to meet, thrash out their differences, and act. Performance was measured constantly, and rewarded not with cash but with stock in the fast-growing company.

Hire good people, create a culture of healthy debate, reward with stock to align incentives to make sure people disagree and commit instead of endlessly play politics.

Above all, he kept his eye on the money. Teletypes clattered constantly as outlying terminals sent booking information to headquarters. Clerks updated records showing how many days each container had carried revenue traffic, how many tons it had hauled, how many dollars it had grossed. Geographic analysis documented the land transportation patterns of Sea-Land cargo. Monthly financial reports revealed how much revenue Sea-Land received from each commodity shipped from Newark to Texas, reminding all that an eighteen-ton container of liquor was twice as profitable as a four-ton container of toys. Weekly reports documented cash flow.

And there was an endless stream of demands for better cost control. Shaving 1.6 cents off the cost of handling 100 pounds in Ponce could save $14,300 a year. One more container lift per gang hour would save $180,000. Limiting long-distance telephone calls to three minutes would save $65,000. "Probably more attention was paid to financial results there than you find in any company around today," remembered Earl Hall, later Sea-Land's chief financial officer. In 1961, its sixth year, Sea-Land's container business had finally moved into the black. So long as McLean was involved in running it, Sea-Land never lost money again.

I've had business/product managers tell me that metrics are easy to game. Make revenue a metric and have your people game it. It's a nice game to win.

One of the ways to win at that game is to control one's expenditure by analysing costs and making small adjustments that aggregate into big savings overall.

Sea-Land's practice was to write off its ships over six years, an unusually short period for long-lived assets. Very high write-offs made the short-term profit picture look bleak, but it meant that Sea-Land could report very high profits a few years later, once the ships had been fully depreciated. This accounting, deliberately designed to depress short-term profitability, was not widely appreciated by analysts who examined the company's financial reports. In the mid-1960s, the Internal Revenue Service forced Sea-Land to depreciate its ships over fifteen years instead of six, and its financial reporting became less obscure.

McLean's hasty capital deprecation strategy must have put managers under more pressure to deliver in the short term and simultaneously ensured higher profits in the future.

They are turning this job into a factory job complained New York longshoreman Peter Bell. Agreed Sidney Roger, editor of the ILWU newspaper in San Francisco, "I've heard so many men say, this is exactly what they said: 'It's no fun any longer working on the waterfront. The fun is gone.' The fun has to do with the men working together, a sense of camaraderie"

Made me mindful about the immediate impacts of automation. People are used to their routine and derive a sense of purpose feel a sense of loss, even when automation removes the aspects of physical labour.

Despite these discontents, the longshore unions' tenacious resistance to automation appeared to establish the principle that long-term workers deserved to be treated humanely as businesses embraced innovations that would eliminate their jobs. That principle was ultimately accepted in very few parts of the American economy and was never codified in law. Years of bargaining by two very different union leaders made the longshore industry a rare exception, in which employers that profited from automation were forced to share the benefits with the individuals whose work was automated away.

It's possible to amicably resolve the next wave of industrial automation if the experience of longshoremen is taken into account.

Forgash impelled his committee to reach consensus quickly. By late summer of 1959, it had agreed unanimously that "standard* containers would be 20 feet or 40 feet long, 8 feet wide, and 8 foot high. No matter: individual companies preferences, Forgash asserted, would have to yield to the need for uniformity. "Even if we reach the goal slowly, we must have a goal" he said. "Otherwise, obsolescence will overtake us all if each man is his own engineer."

Suboptimal uniformity sooner is better than allowing entropy. The formatting by gofmt isn't anyone's favourite, but it's better than 1000s of developers each developing their own style like C++.

Four of the leading steamship lines, Sea-Land, Matson, Alaska Steamship, and American President line, fought back, because adoption of the National Casting fitting would have required them to change all of their containers. Instead, they proposed a minor change to the fitting that the MH-5 committee was designing based on the Sea-Land patent. If the hole on the top of the fitting were moved by half an inch, they estimated, 10,000 containers – about 80 percent of all large containers used by U.S. railroads and ship lines other than Sea-Land-would be "reasonably compatible" with Sea-Land's.

One of the advantage of being an early adopter is the ability to steer standards that will last for decades if successful. Conversely, this shows that the short-term concerns of the standardization participants can have long-term repercussions.

Through 1966, engineers around the world tested the new fittings and found a variety of shortcomings. As an extra check, a container was put through emergency rests in Detroit, just ahead of another meeting of the iso committee. It failed, the fittings on the bottom of the test container giving way under heavy loads. When TC104 convened in London in January 1967, it was faced with the uncomfortable fact that the corner fittings it had approved in 1965 were deficient. Nine engineers were named to an ad hoc panel and told to solve the problems quickly.

End-to-end test your standard with real-world inputs. Having participated in fundamental library design, I have seen how important it is to integrate the new proposed API into an existing system and watch how it handles real traffic. I also expect the real-world to come up with unexpected use cases, which encourages me to seek contact with reality as early as possible.

For all of their earthy bluster, their businesses had survived thanks almost entirely to government coddling. On domestic routes, government policy discouraged competition among ship lines. On international routes, rates for every commodity were fixed by conferences, a polite term for cartels, and the most important cargo, military freight, was handed out among U.S.-flag carriers without the nuisance of competitive bidding. Decisions about buying, building, or selling ships, about leasing terminals, and about sailing new routes all depended on government directives. For men who had prospered in this environment, who loved the smells of the ocean and fondly referred to their ships as "she," Malcom McLean's wholly unromantic interest in moving freight in boxes had little appeal.

I now question if people who romanticise about the way things have always been are just worried about progress.

without the nuisance of comptetitive bidding

Similar to the pre-SpaceX time of satellite and defence contracts in the US. NASA was a monopolist with no incentive to even try 10x improvements.

the collective wisdom of the shipping industry held that they would never carry more than a tenth of the nation's foreign trade

Their romanticism blinds them to the long-term effects and makes them bad predictors.

Liquor exporters had long complained of huge losses to theft on the docks, and convincing them to use containers was not a hard sell. Among Sea-Land's first ports of call was Grangemouth, in Scotland, where it picked up Scotch whiskey. Sea-Land won the business with a tank container, made of stainless steel, designed to let exporters ship their whiskey in bulk for bottling in the United States. Two tank containers would fit neatly into a standard container cell on a Sea-Land ship, putting an end to the pilferage that had plagued the whiskey trade

I think this was an accidental by-product of containerization, not an explicit use case taken into consideration when the shipping container and dock infrastructure were being designed. I suspect the biz dev types at the time realised that containers are qualitatively better at delivering cargo with minimal risk of theft and must have targeted companies whose goods were often stolen from breakbulk storage. Other types of products must have been:

  • canned food
  • fruit and vegetables
  • clothes

The military role was even more crucial. As a U.S.-flag Carrier, Sea-Land was entitled to carry a portion of the freight for the quarter milion U.S. soldiers in West Germany, and the military, determined to push containerisation, channeled cargo Sea-land's way.

I wonder if Flexport has a similar role now.

In March 1966, United States Lines carried the first large containers, along with other freight, on a voyage from New York to London. The following month, Sea-Land's Fairland steamed across the North Atlantic to Rotterdam, Bremen, and the Scottish port of Grangemouth, carrying only containers. With barely a year's notice, Rotterdam and Bremen had lengthened docks, deepened channels, and begun installing container cranes. London had not-and Fairland did not bother to call.

The port of London only has its own slowness to blame. Similar to tech companies that don't rapidly adapt to technological changes.