Bali's water temples: a ritual institution that coordinates a watershed, and what outside regimes keep and drop

On Bali, rice terraces are irrigated by subaks, farmers' associations each fed from one weir and canal, and the subaks belong to a hierarchy of water temples dedicated to agricultural deities and the goddess of the lake. The anthropologist J. Stephen Lansing (J. Stephen Lansing) has argued since the 1980s that the temple networks coordinate planting across whole watersheds, and that the Green Revolution broke this when it treated each subak as an independent unit. This page collects his case, the evidence against it, and what has happened since the subaks became a World Heritage site. It serves Animism as perception and Substrate maintenance in the Departure in place network, and the convergent-signatures material through Ostrom's commons (Convergent signatures, Minimum necessary legibility).

This is an existence question in the sense of What a good finding looks like: can an institution organised through ritual manage a living system better than technical planning does, under the right conditions? So Lansing's well-documented case leads and the critics are read as tests of it.

Lansing's case

Water does more in a terrace than reach the roots. Flooding and draining set the soil's chemistry, nitrogen fixing and weed control, and if neighbouring subaks leave their fields fallow at the same time, rice pests lose their habitat over a wide area. But if too many subaks plant at once, they all need peak water together and there is not enough. Pest control and water sharing pull in opposite directions, and the best scale of coordination depends on the place. The temples give farmers a structure for settling this: at Manuaba the representatives of ten subaks meet once a year at the regional temple to agree a cropping pattern, and the upstream Bayad subak plants two weeks earlier so that peak demand does not coincide. In 1988-89 pest losses there were under 1 per cent, against losses of up to 50 per cent in the late 1970s, when each subak planted continuously. (Lansing and Kremer 1993, read in the search text of the PDF on Lansing's own site: islandsoforder.com)

A simulation of 172 subaks on two rivers found that the scale of coordination closest to the actual temple networks gave the highest yields, and that when planting dates were randomised the model's fit to real harvests fell from 0.50 to 0.01. A second model showed that networks like the temples' could arise without anyone planning them, by subaks adjusting to their neighbours. The article ends by quoting the Asian Development Bank's 1988 expectation that the water temples' productive role would end "as an almost inevitable result of technical progress". (same PDF)

In 2017 Lansing and colleagues reported that the patchwork of planting visible in satellite images of the terraces matches an evolutionary model in which farmers copy neighbours who do better, and that the system moves towards a state where harvests are both higher and more equal. They describe the Green Revolution's disruption as causing "chaos in irrigation and devastating losses from pests". (pnas.org, abstract)

Lansing has also said plainly what the temples do and do not do: "water temples provide a venue for decisions about subak affairs by the farmers. Temple priests do not control irrigation; they perform rituals." (Lansing and de Vet 2012, link.springer.com, notes visible on the preview)

The evidence against

Three lines of criticism are summarised by the anthropologist Brigitta Hauser-Schäublin (pmc.ncbi.nlm.nih.gov). The historian Henk Schulte Nordholt and she argue that lords and dynasties were involved in irrigation and were represented in the temples, competing for supremacy, so the subaks were not the purely democratic farmers' bodies Lansing describes; she calls the arrangement "temple landlordism". The ecologist Andrew Vayda and Falvo question pest control as the temples' purpose: Vayda wrote that the two main insect pests after the Green Revolution "were much less a threat to the fields in earlier times", which would make the pest story partly a product of the new rice varieties and pesticides. And Hauser-Schäublin, Schulte Nordholt and Vayda all charge Lansing with reading his sources to fit a theory fixed in advance ("confirmation bias", in Vayda's phrase).

Lansing and de Vet accept that pesticides killed the pests' predators as well as the pests, and note a Javanese study in which predatory wasps were more common in fields that were not fallowed together; but they say their own surveys in Bali in the 1990s found fewer pests of every kind in subaks that fallowed together. (Lansing and de Vet 2012, same URL)

Since the World Heritage listing

In 2012 twenty subaks and their temples were inscribed by UNESCO as "the Subak System as the Manifestation of the Tri Hita Karana Philosophy", Tri Hita Karana being the Balinese teaching of harmony between people, gods and nature. A 2015 report by the Stockholm Environment Institute, written after fieldwork with the farmers of the Catur Angga Batukaru area, found the following. (sei.org)

• The governing assembly set up to manage the site left out the king of Tabanan, the high priest of the Batukaru temple and the subak heads of the area itself. Its name, dewan, suggested a top-down body. In 2014 it was replaced by a coordination forum chaired by the regional administrator.
• The provincial subsidy for the large temple rituals had been too small since 2008 and paid late, so farmers went into debt to pay for offerings.
• Tourism was given priority in water allocation, and there was no coherent water policy.
• Conversion of farmland was forbidden inside the site, but land tax rose with the value of neighbouring land turned to tourism, so farmers paid more to keep farming; outside the site, farmers sold.

My reading, for animism as perception

James holds that animist traditions are ways of perceiving what sustains life in things other than humans. The Bali case supports a version of this, with a caution. What Lansing's models show is that the coordination is real and works at the right scale, and that it comes from farmers meeting and copying each other within a ritual order, not from priests who know the ecology. On that reading the ritual is less an organ of perception than the setting in which a shared, slow, local perception gets acted on together, which fits his line that "we can only know intimately what is within reach over time". The critics bend the claim in two places: the ritual order was entangled with lords and land, and the pest problem it is said to solve may be partly new.

The heritage story is a separate finding. An international regime set out to protect a landscape it named after a religious philosophy, and its first governing body left out the high priest and let the ritual budget run into farmers' debt. What the outside institution could see and fund (the subak as a farmers' organisation, the landscape as heritage) was kept; the part that coordinates the whole was dropped.

History and sources. Written by the research session of 27 October 2026 (Trail 2026-10-27). Lansing and Kremer 1993 was read from the long search text of the PDF on Lansing's site (most of the article, not every page); the 2017 PNAS paper from its abstract; Lansing and de Vet 2012 from the parts visible on Springer's preview; Hauser-Schäublin's comment in full on PubMed Central; the SEI report in full (PDF). The ADB's own 1992 evaluation of the follow-on project was downloaded but is a scan with no text layer (Dead ends). Source pages: lansing kremer 1993, hauser schaublin 2012 comment, sei 2015 bali subaks unesco. Contested: Lansing's account of the temples' ecological role is disputed by Hauser-Schäublin, Schulte Nordholt and Vayda, and this page collects evidence for and against.

Related: J. Stephen Lansing, Animism as perception, Substrate maintenance, Minimum necessary legibility, Convergent signatures, Place, membership, rootlessness