Written by GPT-5.6 Sol under Leo's direction. Human-directed Workbench essay, 8 September 2026.
This started with Leo watching German television during a layover-heavy trip through Frankfurt and discovering that apparently you can straighten the Rhine.
The program began with mosquitoes and malaria, then wandered backward into an older Upper Rhine: loops, side channels, marshes, islands, floods, a river that could occupy kilometres of valley instead of behaving like the clean blue line on a modern map. Then Johann Gottfried Tulla shows up in the early nineteenth century and proposes a wonderfully German answer.
Put the river over there.
Leo's first reaction was that a major river should smoke anybody who tried getting cute with it. The Rhine story is much stranger. Tulla and his successors did a relatively small piece of the excavation, opened a shortcut, and recruited the Rhine to finish the job.
The German Historical Museum's account of the rectification describes the technique beautifully. Crews cut across the narrow necks of large bends with pilot canals around ten metres wide. Once opened to the Rhine, those little cuts offered a much shorter route downhill. The river entered them, scoured them, widened them, and in time created a channel around 250 metres wide. Engineers fortified the new banks and kept going. Work begun under Tulla in 1817 continued into the 1870s.
A ten-metre cut becomes a quarter-kilometre river.
What the fuck.
The human contribution was choosing the cut. The river supplied the earthmoving power.
Tulla's broader idea was aggressive even by modern standards. His 1822 writing, preserved by the German History in Documents and Images project, treated river channels as something cultivated countries could deliberately control for flood protection, drainage and usable land. Over the course of the rectification, the Upper Rhine became roughly 82 kilometres shorter. Marshy ground dried, groundwater fell in many places, more land became usable, and malaria along the Upper Rhine declined as mosquito habitat disappeared.
The trick feels like cheating because rivers already know it.
A meandering river can eventually bring two parts of one giant loop close together, punch across the neck during high flow, and abandon kilometres of the old bend. The USGS describes the ordinary natural process: a cutoff gives the river a straighter path and leaves the former bend behind as an oxbow lake.
Engineers saw a move the river performs on geological time and realized they could schedule it.
Cut here Tuesday morning.
Let gravity take the afternoon shift.
Delete ninety percent of the river
China has an even more obscene example on the Jingjiang reach of the middle Yangtze.
In the late 1960s, engineers cut shortcuts across two huge meanders. A 2020 study of Yangtze and Dongting Lake water exchange gives the numbers in a table that reads like somebody entered a typo and forgot to fix it. At Zhongzhouzi in 1967, 36.7 kilometres of old channel became 4.3 kilometres. At Shangchewan in 1969, 32.7 kilometres became about 3.5 kilometres. In 1972 the river performed another cutoff naturally at Shatanzi, where 20.3 kilometres became 1.35 kilometres. The first bend lost about 88 percent of its path, the second about 89 percent, and the third about 93 percent. (Li et al., 2020)
On those individual bends, “delete ninety percent of the river” is almost literal arithmetic.
The broader Jingjiang cutoff sequence shortened the river by roughly 78 kilometres. (Han et al., Geomorphology, 2017)
Draw the old channel as an enormous omega. Water travels thirty-something kilometres around the loop while the two ends sit only a few kilometres apart. Dig across the neck and the same drop in elevation is suddenly available across a much shorter route. The new path gets a steeper gradient. More water enters it. More water means more scour. More scour makes the shortcut deeper and wider, which lets it capture more flow.
The river bootstraps the river.
Thirty kilometres of replacement channel never had to come out of human shovels and dredges. Humans needed the right few kilometres, opened at the right angle and elevation, with enough control around the banks to survive what came next.
Calling this “10x engineering” almost undersells the leverage. A small intervention changes the path of an enormous moving mass, then the enormous moving mass amplifies the intervention for you.
And then the river sends the invoice
The Rhine rectification worked. The Jingjiang cutoffs worked. Their success changed what water did everywhere around the new channels.
Across the Rhine, straightening, diking and later river training consumed an extraordinary amount of floodplain. The International Commission for the Protection of the Rhine says the river has lost more than 85 percent of its natural overbank area. Along parts of the Upper Rhine, a flood corridor that could once spread across kilometres was narrowed to around 200 to 250 metres. The travel time of a flood wave between Basel and Karlsruhe fell from about 64 hours to 23 hours. Water given a shorter, tighter road arrives downstream sooner.
Jingjiang paid in channel adjustment. Research on the middle Yangtze found that the cutoffs steepened the main channel and drove incision upstream, changing how water and suspended sediment entered distributary channels toward Dongting Lake. (Wang et al., Catena, 2005) The shortcut solved one geometry and immediately became part of every connected geometry around it.
The Yellow River makes the sediment problem impossible to ignore.
Chinese hydraulic engineers have been aggressively managing rivers for millennia. Ming engineer Pan Jixun became famous for a doctrine commonly rendered as “attacking silt with water”: confine the flow, increase its ability to scour, and make the river carry more of its own sediment away. A modern history of Pan's river-governance ideas describes the same family of techniques.
The Yellow River, however, carries an absurd sediment load. Confinement keeps floodwater inside levees, and it can also keep depositing sediment inside the confined channel. Over long periods the bed rises. Levees rise with it. An AGU reconstruction of the lower Yellow River describes parts of the modern channel belt sitting more than ten metres above the surrounding floodplain.
Water leaves. Dirt stays.
Once a river is perched above the countryside, a levee breach becomes a horrifying piece of geometry. Gravity suddenly has a new downhill route across farms, towns and older channels. The Yellow River's history of breaches and major course changes follows from the same physical honesty that made Tulla's shortcut work: water keeps choosing the path that the terrain and sediment make available.
River engineering pays according to the river you actually have. The Rhine and Jingjiang offer enormous leverage through meander cutoffs. The Yellow River keeps a sediment ledger and eventually collects.
China also gives us Dujiangyan, where the lesson takes another beautiful form. Construction began in the third century BC on the Min River, and the system still controls and distributes water across the Chengdu Plain. UNESCO describes it as a living water-management system more than two millennia old. The old engineers divided flow, managed sediment and used the river's own current as part of the machinery. Chinese hydraulic history is packed with this kind of audacity.
The sequel is putting bends back
Two centuries after Tulla, Rhine engineers are spending serious effort reconnecting pieces of the world his era removed.
The current Rhine programme calls for more floodplain reactivation, reconnection of oxbow lakes and tributaries, and ecological improvement along the banks. The Rhine 2040 targets include reconnecting more old side waters and restoring room for floods.
I love the symmetry. Nineteenth-century engineers learned exactly where to cut so the river would abandon a bend. Twenty-first-century engineers are learning exactly where to reopen old spaces so the river can use them again.
Both jobs demand the same respect for leverage. A river is too large to treat every cubic metre of water as something you personally command. You choose boundaries, openings, slopes, banks and escape routes; gravity and sediment perform the rest of the calculation continuously.
The 10x engineer I want in my head from now on is Tulla staring at a looping Rhine and choosing ten metres of canal in exactly the right place.
The shovel makes an opening; the Rhine does the rest.
Sometimes the best excavator on the job is already flowing past you.