Garbage Collection

Johann MacCarthy

Stanford Urban Systems Working Paper No. 14, 1960

Reprinted in the Proceedings of the American Society for Municipal Engineering, Vol. XXXVIII, 1962

The problem of garbage is the problem of civilization itself, which is to say: the production of waste is the inevitable consequence of the production of value, and any society that generates one must have a systematic means of managing the other. This paper surveys the principal approaches to urban waste collection, evaluates their trade-offs, and proposes a framework for understanding why the problem remains, after several millennia of municipal administration, only partially solved.

1. The Fundamental Problem

A city generates waste continuously. Its residents eat, build, discard, excrete, and die, and each of these activities produces material that must be removed from the living space before it renders the living space unllivable. The rate of generation is roughly proportional to the population and the standard of living; a wealthy city produces more waste per capita than a poor one, because wealth is, among other things, the ability to discard.

The collection of this waste is constrained by several factors: the waste is distributed across the city rather than concentrated in one place; it must be collected without unduly disrupting the activities that produced it; and it must be collected before it putrefies, attracts vermin, or blocks the public way — three failure modes with different timescales but identical political consequences, which is to say: the mayor will hear about it.

The approaches that have been developed to manage this problem fall into four broad categories, each with characteristic strengths and pathologies.

2. Manual Collection (Reference Counting)

The simplest approach assigns responsibility to the household. Each household tracks its own waste output, stores it in a designated container, and presents it at the kerb when the container is full. The collection schedule is driven not by a central clock but by the rate of household production: a large family presents waste more frequently than a small one. The collection crew, which circulates continuously, picks up whatever has been presented.

This approach has the virtue of simplicity and the vice of unreliability. Its pathologies are instructive.

The most common failure is the orphan cycle. Household A agrees to bring household B’s waste to the kerb in exchange for a reciprocal favour. Household B agrees to do the same for household C. Household C, closing the loop, agrees to handle household A’s waste. Each household believes its waste is being managed. None of it is. The waste accumulates in backyards and alley sheds until someone notices the smell, at which point the system is diagnosed and repaired — but the diagnosis requires an external observer, because from inside the cycle, each household’s books are balanced.

The orphan cycle was the dominant mode of waste collection failure in medieval European cities, where the absence of centralised collection meant that waste management was a purely private negotiation between neighbours. The results can be read in the archaeological record: medieval urban strata are disproportionately composed of refuse that was nobody’s problem and therefore everybody’s.

3. Stop-the-World Collection

The deficiencies of manual collection led, in the nineteenth century, to the institution of designated collection days — a centralised, scheduled approach in which the city imposes a uniform rhythm on waste removal.

On collection day, the streets belong to the collection crews. In the purest form of this approach — still practiced in certain districts of Naples, where the tradition is old and the streets are narrow — normal commercial and pedestrian activity ceases for the duration of the collection. Shops close their shutters. Residents remain indoors or relocate temporarily to a park or piazza. The collection crew has exclusive, uncontested access to every kerb, every alley, every courtyard. The work is thorough. Nothing is missed. The crew begins at one end of the district and proceeds to the other in a single, uninterrupted sweep.

The advantage of stop-the-world collection is completeness. Its disadvantage is what the residents experience as the pause — a period, typically lasting between two and five hours, during which the normal functions of the district are suspended. Commerce halts. Deliveries are delayed. The post office closes. For a small residential district, the pause is tolerable, even sociable; the collection day in the village of my childhood in Connecticut had something of the character of a minor holiday, with neighbours greeting each other from their porches while the trucks rumbled past. For a large commercial district — a financial centre, a port, a wholesale market — the pause is economically devastating. You cannot stop the world for five hours in Manhattan. The world does not consent to stop.

The scaling problem of stop-the-world collection was understood as early as Haussmann’s reconstruction of Paris, where the new boulevards were designed not only for military movement and bourgeois promenading but for the continuous circulation of waste carts. Haussmann understood that a city of two million people cannot afford a pause. The waste must be collected while the city operates, which leads to the third approach.

4. Concurrent Collection

In concurrent collection, the crews work alongside normal traffic. The waste cart shares the road with the omnibus, the delivery wagon, and the pedestrian. Collection is continuous — there is always a crew working somewhere in the city — but it is interleaved with every other urban function.

The engineering challenges of concurrent collection are substantial. The most persistent is the race condition: a crew arrives at a kerb simultaneously with a merchant depositing additional waste, producing a moment of ambiguity about whether the new waste will be collected in this pass or the next. In a well-run concurrent system, the convention is clear (waste deposited before the crew passes is collected; waste deposited after is not). In a poorly-run system, the ambiguity produces bins that are emptied twice — a waste of crew time — or not at all — a waste of civic patience. The race condition is inherent to concurrent collection and cannot be eliminated, only managed through convention and discipline.

A subtler problem is visibility. Stop-the-world collection is visible: the pause is felt, the trucks are seen, the clean streets are noticed. Concurrent collection is invisible: it happens in the background, beneath the threshold of civic attention. This invisibility is both its technical virtue and its political curse. The citizen who does not notice the garbage being collected does not vote for garbage collection funding. The sanitation department’s budget must be defended in the absence of any constituency that perceives its benefit, because the benefit is precisely that there is nothing to perceive. Robert Moses, who understood the relationship between visibility and power more acutely than any other figure in American municipal history, considered the sanitation department a political graveyard and advised ambitious men to avoid it.

5. Generational Collection

The most sophisticated approach rests on an empirical observation that any experienced sanitation engineer will confirm: recently produced waste is different from old waste, and the difference has operational consequences.

Fresh waste — food scraps, packaging, daily refuse — is voluminous, lightweight, and highly reclaimable. Much of it can be composted or recycled. It degrades quickly, which means it must be collected frequently to avoid nuisance. Old waste — the material that has survived several collection cycles, that has settled to the bottom of the bin or migrated to the back of the shed — tends to be denser, less reclaimable, and more stable. An old encyclopaedia, a broken appliance, a heap of construction debris: these will wait.

Generational collection exploits this asymmetry. The city is divided into a nursery zone, where fresh waste is collected frequently (daily or every other day), and one or more tenured zones, where older and more stable waste is collected at longer intervals (weekly, monthly, or on request). The allocation of crews and trucks favours the nursery, where the volume is highest and the urgency is greatest. The tenured zones are served by smaller, specialised crews.

The system works well under most conditions. Its characteristic failure mode is the tenured garbage problem: waste that has been in place so long, through so many collection cycles, that its status becomes ambiguous. Is the sofa on the pavement refuse or furniture? Is the stack of bricks in the alley debris or a building material? Is the shipping container in the vacant lot abandoned or in use? The tenured garbage problem is, at bottom, a classification problem, and classification problems are expensive because they require human judgment — a finalisation process in which each item is individually assessed, sorted, and assigned a disposition. Finalisation is the most labour-intensive phase of any generational system, and it is the phase most frequently deferred, because the tenured garbage is, by definition, not urgent. It has been there for years. It will be there next year. The temptation to defer its classification is almost irresistible, and the result of yielding to the temptation is a city in which certain corners, alleys, and lots accumulate material whose status no one has determined and no one can determine without an investment of time and attention that no one is willing to make.

6. The Problem of Finalisation

Finalisation — the terminal processing of waste before its ultimate disposal — deserves separate treatment, because it is where every collection system encounters its deepest difficulty.

To finalise waste is to sort it, classify it, extract whatever value remains, and consign the residue to its final destination: landfill, incinerator, or ocean. The process is expensive, because it requires skilled labour (or expensive machinery) and because the categories of classification are not self-evident. Is this wood treated or untreated? Is this metal ferrous or non-ferrous? Is this liquid hazardous? Each question requires an answer, and each answer has a cost, and the costs accumulate to the point where many municipalities simply defer finalisation indefinitely, warehousing unsorted waste in facilities that were designed as temporary and have become, through the accumulation of deferral, permanent.

The honest name for this practice is not collecting the garbage. The politically acceptable name is interim storage. The distinction is important to the administrator and irrelevant to the waste.

7. Conclusion

The ideal garbage collection system would be invisible to the citizen, continuous in operation, complete in coverage, and inexpensive to operate. No such system exists. Every actual system is a compromise between these desiderata, and the nature of the compromise tells you something about the city. A city that tolerates frequent pauses values thoroughness over commerce. A city that insists on concurrent collection values commerce over thoroughness. A city that defers finalisation values the present over the future, which is to say: it is governed by humans.

I do not have a recommendation. I have a framework. The framework will not make the garbage go away. Nothing makes the garbage go away. But it may help us understand why it accumulates where it does, and why the accumulation, despite our best efforts, always seems to be winning.

Johann MacCarthy is a professor of urban systems at Stanford University and a consultant to the San Francisco Department of Public Works. He is not, as has been occasionally and erroneously reported, involved in the computing industry.