a proposed development philosophy

SESERAGI AGILE A water-cycle model for teams that never run dry

A waterfall falls once, and never rises again.
A stagnant pool eventually rots.
The Genbei River has kept flowing for nearly a century without drying up, by pumping its own downstream spring water back upstream in a different form.
This structure, mapped onto the rhythm of development.

🌊 Not a rival to Waterfall, but its missing complement 💧 Where Specification-Driven Development meets Agile 📜 A proposed philosophy, in the lineage of KAIZEN

Why Talk About a River, of All Things

Waterfall and Agile are always framed as opposites. But as water metaphors, both of them are actually incomplete.

WATERFALL

A Waterfall — Once It Falls, It Never Climbs Back

  • Requirements → design → implementation → testing → release: water flows one way, from high ground to low.
  • Peer down into the pool from upstream (management, planning), and all you see is spray — the higher you stand, the less you can actually see of what's happening on the ground.
  • Problems found downstream (production, users) cost too much to carry back upstream (design) — in practice, they never make the trip.
  • The water is used, falls, and that's the end of it.
THE USUAL "AGILE"

A Stagnant Pool — Stirred, But Never Drained

  • Repeating the sprint becomes the goal itself, while actually releasing into the river outside — production, customers — keeps getting pushed back. The surface stays busy, churned, going nowhere.
  • "Retrospectives" and "kaizen" stay sealed inside the same team, the same org — a textbook case of local optimization.
  • It repeats, but no new water — real production data, the customer's actual reality — ever flows in from outside. It quietly stagnates.
  • Meanwhile the boss is on the bank, fishing — watching not the health of the whole pond, but only his own catch: this quarter's KPIs.
SESERAGI AGILE (PROPOSED)

A Seseragi (Murmuring Stream) — Pumped Up, Reshaped, Flowing Again

  • "Water" drawn from downstream (production, real usage) is deliberately pumped back upstream (planning, design).
  • But it's not the same water simply going in circles. Each time, it's reshaped to serve a new purpose before being passed on (see the Principle of Cascading Reuse, below).
  • Neither a violent waterfall nor a sealed pond — it holds the rhythm of a small stream, flowing on, comfortably, without end.
SESERAGI AGILE
noun · development philosophy / seseragi (murmuring stream) + agile

A way of developing software that neither dumps its outputs and feedback downstream just once, like a waterfall, nor merely circulates them inside one closed team or org. Instead, it "deliberately draws from downstream — where the work is actually used — and lets that value keep flowing back upstream, reshaped, into planning and design," so that neither the team nor the product ever runs dry, while sustaining a small, repeating current.

It draws its inspiration from the Genbei River in Mishima, Shizuoka — a flow that, left to nature, would have dried up, but has instead been kept alive for nearly a century through deliberate human "pumping" and "passing on."

The Genbei River, as Primary Source

"Pumping water from downstream back upstream" turns out, once you check the details, to be a more specific structure than that. To test the idea, we went to primary sources — government agencies, the company involved, and the local NPO.

1

Spring Water — the Original Source, Nearly Dried Up

The Genbei River's original source was groundwater from Mt. Fuji, welling up at the tip of the Mishima lava flow (Rakujuen Park's Kobamaike Pond). Urbanization and rising groundwater use from the late 1950s onward lowered the water table, and by around 1962 the pond's spring had effectively stopped flowing — the river became little more than an open sewer.

2

The Pump — From the "Downstream End" of the Same Aquifer

Not downstream of the Genbei River itself — the source is spring water from the Kakita River, downstream within the same Mt. Fuji aquifer system, pumped and piped in by a prefectural industrial water utility.

3

First Use — A Change of Role

Toray's Mishima plant, north of the station, buys this water as industrial water and uses it once as primary cooling water — for temperature control only, never in contact with chemicals. The water quality is untouched; only its role shifts, from "spring water" to "clean, used water."

4

Return — Back to the Very Top

The used water travels through existing drainage pipes and is released from Rakujuen Park, at the Genbei River's furthest upstream point. This began on January 30, 1992. The flow has grown from an initial 700 m³/h in winter to today's 1,500 m³/h in summer and 900 m³/h in winter.

5

Passed On Again — Downstream to Irrigation

Having flowed through the Genbei River, the water is warmed by sunlight at the downstream Nakago warming pond, reused once more as irrigation water, and finally joins the Kano River system. One flow of water carries its value forward — industrial, then environmental and recreational, then agricultural — changing purpose each time.

6

The Broker — Who Designed the Flow

In 1992, eight citizen groups came together to found Groundwork Mishima. Mediating between government, industry, and residents — three parties with different interests — it delivered Japan's first "Groundwork"-method redevelopment, restoring public access to water across eight zones. The result: the return of the Hotokedojo loach, Mishima baikamo water crowfoot, and Genji fireflies; designation as a World Heritage Irrigation Structure in 2016; and a Good Design Gold Award in 2025.

To Be Precise: this isn't a closed "cycle." It's a multi-stage cascade of reuse: the same water isn't circling the same loop, but being "passed on," changing purpose once, in one direction — spring → industry → environment → agriculture. The Ministry of the Environment's own materials also note that "there is no guarantee this discharge can continue indefinitely" — worth noting that this system rests on goodwill and agreements between parties, and carries that fragility with it.

The Four Principles of Seseragi Agile

Mapping the Genbei River's structure directly onto principles for a development team.

1

Left Alone, It Dries Up

Feedback doesn't arise on its own. Just as Kobamaike Pond dried up silently, user feedback, operational data, and incident reports never make it back upstream unless you deliberately build the "pump" — instrumentation, analytics, regular user interviews.

2

Never Spend It All at Once — Reshape It, Then Pass It On

What a retrospective or kaizen session learns shouldn't be consumed and forgotten within that one team's sprint. Just as cooling water becomes water for public amenity, then irrigation water, learnings get documented and abstracted, and passed on as "irrigation water" for other teams and future projects.

3

A Flow Doesn't Exist Unless Someone Designs It

Just as Groundwork Mishima stood between government, industry, and residents, someone has to stand between leadership, engineering, and users — a broker (a PM, a Scrum Master, a platform team) who lays channels across competing interests. If every department or team chases only its own interest, the water never completes its cycle at all.

4

Not a Torrent, Like a Waterfall

Drop a huge volume of data or requirements all at once, and downstream — the team — floods. What makes a seseragi pleasant is a steady, unforced flow that never stops. Sustainable pace is a question of rhythm, not volume.

The Foundation of Whole-System Optimization — the Toyota Production System

KAIZEN is often introduced abroad as "an attitude of continuous small improvement." But the real aim of the Toyota Production System is not the local optimization of a single process or team — it's whole-system optimization, applied to the entire flow of material and information. SESERAGI AGILE is also an attempt to translate that abstract goal into the concrete structure of a river.

JIT

The Right Water, at the Right Time, in the Right Amount

The Genbei River's flow is adjusted by season (1,500 m³/h in summer, 900 m³/h in winter). Too much floods it; too little dries it out. Overproduction — muda — is the greatest waste there is, in water and in software alike.

Jidoka

When Something Goes Wrong, Stop the Flow and Say So

In 2017, the Genbei River's water level suddenly dropped — not because the discharge had stopped, but because of a misadjusted upstream gate, fixed the very next day. A working example of a system that detects an abnormality, halts, and gets fixed.

Next Process

Treat What's Downstream as a Customer

The cooling water Toray uses affects the ecosystem of the Genbei River downstream, and the people who use it as irrigation water further downstream still. Without treating the next process as a customer, no one would bother with water quality or temperature at all.

Whole System

Redesigning Individual Interests Into a Whole

If government, industry, and residents had each looked only at their own interest, this return-flow structure would never have existed. It exists only because Groundwork Mishima, as broker, redesigned the region's entire water system as a single system.

What Was Already Flowing Into the Agile Manifesto

SESERAGI AGILE doesn't import a new philosophy from outside. It re-visualizes, through the river metaphor, something that had already flowed into the Agile Manifesto (2001) itself — Japanese-born lean/TPS thinking.

The 17 people who drafted the Agile Manifesto brought together multiple streams — XP, Scrum, DSDM, Crystal, and more. It isn't a simple, linear story of "they studied TPS and wrote the Manifesto." But lean thinking flowed in through at least two tributaries.

One is the paper Jeff Sutherland and others drew on when building Scrum in the 1990s: Ikujiro Nonaka and Hirotaka Takeuchi's "The New New Product Development Game" (Harvard Business Review, 1986). It studied new-product teams at Fuji Xerox, Canon, Honda, NEC, Epson, Brother, 3M, and HP — not Toyota itself — and its rugby-scrum metaphor is where the name "Scrum" comes from, directly.

The other is Mary and Tom Poppendieck's Lean Software Development (2003), which translated Toyota Production System thinking directly for software and was widely read across the Agile world. The 2020 Scrum Guide states outright, in its opening lines, that "Scrum is founded on empiricism and lean thinking." In other words, lean thinking wasn't invented for Agile — it flowed in, from more than one tributary, converging into it. That's a shape remarkably similar to the cascade structure Seseragi Agile describes.

Agile Manifesto (official text)Lean / TPS CounterpartSESERAGI AGILE Principle
Value: "Customer collaboration over contract negotiation" The next process is the customer Principle of the Broker — laying a channel of cooperation across competing interests
Principle 8: "...promote sustainable development... maintain a constant pace indefinitely" Heijunka (leveling, eliminating mura) Principle of Sustainable Flow — never drying up, never overflowing
Principle 10: "Simplicity — the art of maximizing the amount of work not done — is essential" Elimination of muda Principle of Cascading Reuse — not a single drop wasted
Principle 12: "At regular intervals... tunes and adjusts its behavior accordingly" Kaizen Principle of Deliberate Return — a retrospective is the pump that returns downstream water upstream
To Be Precise: this isn't a claim that "TPS produced the Agile Manifesto." There's no record of the 17 drafters directly citing lean/TPS as an intellectual source, and agilemanifesto.org's own account of its history mentions neither Toyota nor Nonaka and Takeuchi. What can be shown is narrower: that lean thinking flowed into Agile through one specific tributary — Scrum — and was made explicit in the 2020 Scrum Guide.

A Footnote: Resonance With the SECI Model

Ikujiro Nonaka and Hirotaka Takeuchi, authors of the 1986 paper that gave Scrum its name, went on to propose the SECI model nine years later in The Knowledge-Creating Company (1995). Scrum co-creator Jeff Sutherland, in his talk materials "The Roots of Scrum" (2005), calls the two of them "Godfathers of Scrum" and lists this book among his references, too.

The essence of the SECI model isn't its four mechanical steps of converting tacit knowledge into explicit and back. Knowledge isn't consumed by use — every round of dialogue and practice widens the circle of people involved (individual → team → organization → across organizations), and the value itself amplifies. That "spiral-up" is the real core of it.

The Genbei River's water traces the same shape. Spring water is a source of tacit knowledge, in a sense — the gut instinct and experience of a place, not yet put into words. Had it simply ended as industrial water for Toray, that would have been mere consumption. What actually happened is that, through the "dialogue space" that was Groundwork Mishima, one company's cooling water became a public waterside space, the Hotokedojo loach and Genji fireflies returned, and the circle of people and recognition kept widening — a Top 100 Waterway (2006), a Top 100 Natural Water Source (2008), a World Heritage Irrigation Structure (2016), a Good Design Gold Award (2025). That's remarkably close to the SECI spiral: tacit knowledge becoming explicit through a "ba" (shared place), systematized as organizational knowledge, taking root in many people through practice, and from there giving rise to new tacit insight all over again.

The Core of SECISeseragi Agile
Knowledge isn't consumed — dialogue and practice amplify it Cascading Reuse — the more it's used, the more value it hands forward
Spiral-up — from individual to group to organization to inter-organization, widening the circle 30 years of widening the circle: one company's cooling water → a public waterside space → world recognition
Ba — a shared context deliberately provided for knowledge conversion (Nonaka & Konno, 1998) The Broker — Groundwork Mishima built the space for dialogue
Internalization — through practice, explicit knowledge takes root again as new tacit knowledge A retrospective's lessons quietly becoming "how this team just does things"
To Be Precise: SECI is a theory of organizational knowledge creation, not a description of water cycles as such. The mapping above stays at the level of a conceptual analogy. Strictly speaking, it isn't the water molecules that trace the spiral-up — it's the practice and dialogue of the people who have engaged with this river for more than 30 years. The water is only a visible medium for that accumulating practice.

The Logic That Connects to Specification-Driven Development (SDD)

"Specification-Driven Development" — the idea that the spec itself is the primary artifact, and code is just what an AI agent generates from it, each time — turns out to have almost exactly the same shape as the Genbei River.

Specification
The uppermost source (Kobamaike Pond)
AI-generated implementation
First use (industrial cooling)
Running, testing, production
Downstream spring water (Kakita River)
Checking for drift
The pumping station
Updating the spec
Discharge at the very top

In SDD, the spec plays the role of Kobamaike Pond — the uppermost source. But just as that pond quietly dried up amid urbanization, a spec that's written once and left alone dries up too, in the growing gap with reality. Because AI agents generate implementations from specs so fast, the risk of the implementation and production behavior quietly drifting away from the spec grows just as fast.

That's exactly why you need a mechanism that deliberately pumps the "downstream spring water" — execution results, failing tests, production behavior — back up into the spec itself. Just as Toray doesn't discharge Kakita River water untouched but uses it as cooling water first, execution results only become clean water fit to nourish the spec after passing through one buffer: verification. Pasting raw logs straight into a spec is a flood, not a seseragi. Only with that return flow — spec → implementation → execution → verification → spec update — can SDD escape being just another "requirements doc, written once and done."

Beyond KAIZEN, Toward SESERAGI

KAIZEN, properly understood, was never about the local optimization of one process or team — it's a philosophy of whole-system optimization, applied to the entire flow of materials and information. SESERAGI AGILE is best understood as a construction line for visualizing and designing that same abstract goal as the concrete structure of a river. Not a rival concept, but one standing on the same ground.

DimensionKAIZEN (TPS's whole-system philosophy)SESERAGI AGILE (its concrete form)
Goal Whole-system optimization — not one process's utilization, but the lead time and throughput of the entire flow The same whole-system optimization — made visible as the concrete flow "from spring water to final reuse"
Criteria Eliminating muda, mura, and muri (overproduction, inventory, waiting, transport, over-processing, motion, defects/rework) The four principles: deliberate return, cascading reuse, the broker, sustainable flow
Typical failure The trap of local optimization — efficiency in one process or team creates waste elsewhere: inventory, stagnation No return-flow mechanism, so upstream quietly drifts from reality inside a closed team or org
The key question "Does this improvement actually shorten the lead time of the whole?" "Whose 'next water' does this learning become, once it crosses the org's boundary?"

Never Stop It. Never Flood It. Never Let It Dry.

SESERAGI AGILE isn't an established methodology — it's a proposal, a thought experiment.
Where is the "pump" and the "broker" that would fit your team?

A seseragi never stagnates, no matter how many times it repeats.

Sources & References (fact-checking on the Genbei River)

  1. Ministry of the Environment, "Case Studies on Regional Water Environment Improvement" No.31, Genbei River (PDF) — env.go.jp/water/junkan/case2/pdf/31.pdf
  2. Ministry of the Environment, Spring Water Conservation Guideline, Case 19, "Kobamaike Pond and Others" (PDF) — env.go.jp/water/yusui/guideline/10.pdf
  3. Ministry of the Environment, "Heisei Top 100 Natural Water Sources," Genbei River — water-pub.env.go.jp
  4. NPO Groundwork Mishima, "Genbei River," "Has the Water Disappeared From the Genbei River?" (2017) — gwmishima.jp
  5. Japan Chemical Fibers Association, SDGs case study (Toray), "Cooling Water Discharge Into the Genbei River and Cooperation on Waterside Environment Improvement" — jcfa.gr.jp
  6. Shizuoka Prefecture official site, "Genbei River" / Shizuoka Prefecture Enterprise Bureau, "About the Industrial Water Supply" — pref.shizuoka.jp
  7. Izu Peninsula Geopark, "Rakujuen Park and the Genbei River" / Izu Geo Heritage, "Genbei River" — izugeopark.org, izugeo13.sakuraweb.com
  8. Fujiclean Co., "The Revival of Mishima's Water Scenery" — fujiclean.co.jp
  9. Toray Industries recruiting site, "Mishima Plant" (established 1958) — toray.co.jp

Descriptions of the Toyota Production System draw on Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production (1978), and general explanatory content from Toyota's official site.

The Agile Manifesto and its 12 principles are quoted from the official text at agilemanifesto.org. Ikujiro Nonaka and Hirotaka Takeuchi's "The New New Product Development Game" (Harvard Business Review, 1986), Mary and Tom Poppendieck's Lean Software Development (2003), and the 2020 Scrum Guide are likewise drawn from their original texts and public materials.

Descriptions of the SECI model draw on Ikujiro Nonaka and Hirotaka Takeuchi, The Knowledge-Creating Company (Oxford University Press, 1995), Nonaka & Konno, "The Concept of 'Ba'" (California Management Review, 1998), Jeff Sutherland, "The Roots of Scrum" (JAOO, 2005 talk materials), and Kenji Hiranabe and Ikujiro Nonaka, Agile Development and Scrum (Shoeisha).

"SESERAGI AGILE," as presented on this page, does not refer to any established methodology, certification, or organization — it's a personal thought experiment built from facts about the Genbei River. The descriptions of the river follow the sources above, though sources vary in how specifically they describe certain points (e.g., the primary cause of Kobamaike Pond's drying), and this page defers to the more cautious accounts from government and corporate primary sources where they differ.