places to intervene in a system - the donella meadows ... · 1 folks who do systems analysis have a...

Leverage PointsPlaces to Intervene in aSystem

by Donella Meadows

The Sustainability Institute • 3 Linden Road • Hartland VT 05048Phone 802-436-1277 • FAX 802-436-1281 • Email: [email protected] additional copies of this report, send US$10 to The Sustainability Institute.

Published by

© The Sustainability Institute, 1999.A shorter version of this paper appeared in Whole Earth, winter 1997.

1

Folks who do systems analysis have a greatbelief in “leverage points.” These are placeswithin a complex system (a corporation, aneconomy, a living body, a city, an ecosys-tem) where a small shift in one thing canproduce big changes in everything.

This idea is not unique to systems analy-sis—it’s embedded in legend. The silver bul-let, the trimtab, the miracle cure, the secretpassage, the magic password, the single heroor villain who turns the tide of history. Thenearly effortless way to cut through or leapover huge obstacles. We not only want tobelieve that there are leverage points, wewant to know where they are and how toget our hands on them. Leverage points arepoints of power.

The systems analysis community has alot of lore about leverage points. Those of uswho were trained by the great Jay Forresterat MIT have all absorbed one of his favoritestories. “People know intuitively whereleverage points are,” he says. “Time after timeI’ve done an analysis of a company, and I’vefigured out a leverage point—in inventorypolicy, maybe, or in the relationship betweensales force and productive force, or in per-sonnel policy. Then I’ve gone to the com-pany and discovered that there’s already a lotof attention to that point. Everyone is tryingvery hard to push it in the wrong direction!”

Leverage Points:Places to Intervene in a System

by Donella H. Meadows

The classic example of that backwardintuition was my own introduction to sys-tems analysis, Forrester’s world model.Asked by the Club of Rome to show howmajor global problems—poverty and hun-ger, environmental destruction, resourcedepletion, urban deterioration, unemploy-ment—are related and how they might besolved, Forrester made a computer modeland came out with a clear leverage point:Growth.1 Not only population growth, buteconomic growth. Growth has costs as wellas benefits, but we typically don’t count thecosts—among which are poverty and hun-ger, environmental destruction, and so on—the whole list of problems we are trying tosolve with growth! What is needed is muchslower growth, and in some cases no growthor negative growth.

The world’s leaders are correctly fixatedon economic growth as the answer to virtu-ally all problems, but they’re pushing withall their might in the wrong direction.

Another of Forrester’s classics was hisurban dynamics study, published in 1969,which demonstrated that subsidized low-income housing is a leverage point.2 Theless of it there is, the better off the city is—even the low-income folks in the city. Thatis because subsidized housing withoutequivalent effort at job creation for the

1 J.W. Forrester, WorldDynamics. Portland,Oreg.: Productivity Press,1971.

2 J.W. Forrester, UrbanDynamics. Portland,Oreg.: Productivity Press,1969.

Leverage Points: Places to Intervene in a System

2

inhabitants severely disrupts a city’s employ-ment/housing ratio, effectively increasingunemployment and welfare costs anddespair. This model came out at a time whennational policy dictated massive low-incomehousing projects. Forrester was derided.Now those projects are being torn down incity after city. Forrester was right.

Counterintuitive. That’s Forrester’s wordto describe complex systems. Leveragepoints are not intuitive. Or if they are, weintuitively use them backward, systemati-cally worsening whatever problems we aretrying to solve.

The systems analysts I know have comeup with no quick or easy formulas for find-ing leverage points. When we study a sys-tem, we usually learn where leverage pointsare. But a new system we’ve never encoun-tered? Well, our counterintuitions aren’t thatwell developed. Give us a few months oryears to do some computer modeling andwe’ll figure it out. And we know from bitterexperience that, because of counter-intuitiveness, when we do discover thesystem’s leverage points, hardly anybody willbelieve us.

Very frustrating, especially for those of uswho yearn not just to understand complexsystems, but to make the world work better.

So one day, I was sitting in a meetingabout how to make the world work better—actually it was a meeting about how the newglobal trade regime, NAFTA and GATTand the World Trade Organization, is likelyto make the world work worse. The more Ilistened, the more I began to simmer in-side. “This is a huge new system people areinventing!” I said to myself. “They haven’t

the slightest idea how this complex struc-ture will behave,” myself said back to me.“It’s almost certainly an example of crank-ing the system in the wrong direction—it’saimed at growth, growth at any price!! Andthe control measures these nice, liberal folksare talking about to combat it—smallparameter adjustments, weak negative feed-back loops—are way too puny!!!”

Suddenly, without quite knowing whatwas happening, I got up, marched to the flipchart, tossed over to a clean page, and wrote:

Places to Intervene in a System

(in increasing order of effectiveness)

9. Constants, parameters, numbers (subsi-

dies, taxes, standards)

8. Regulating negative feedback loops

7. Driving positive feedback loops

6. Material flows and nodes of material

intersection

5. Information flows

4. The rules of the system (incentives,

punishments, constraints)

3. The distribution of power over the rules of

the system

2. The goals of the system

1. The mindset or paradigm out of which the

system—its goals, power structure, rules,

its culture—arises.

3

Sustainability Institute, December, 1999

Everyone in the meeting blinked in sur-prise, including me. “That’s brilliant!” some-one breathed. “Huh?” said someone else.

I realized that I had a lot of explainingto do.

I also had a lot of thinking to do. Aswith most of the stuff that comes to me inboil-over mode, this list was not exactlytightly reasoned. As I began to share it withothers, especially with systems analysts whohad their own lists, and with activists whowanted to put the list to immediate use,questions and comments came back thatcaused me to rethink, add and delete items,change the order, add caveats.

In a minute I’ll go through the list thatI ended up with, explain the jargon, andgive examples and exceptions. The reasonfor this introduction is to place the list in acontext of humility and to leave room forevolution. What bubbled up in me that daywas distilled from decades of rigorous analy-sis of many different kinds of systems doneby many smart people. But complex systemsare, well, complex. It’s dangerous to gener-alize about them.

So, what you are about to read is a workin progress. It’s not a simple, sure-fire recipefor finding leverage points. Rather, it’s aninvitation to think more broadly about themany ways there might be to get systems tochange.

Here, in the light of a cooler dawn, is arevised list:

Places to Intervene in a System

(in increasing order of effectiveness)

12. Constants, parameters, numbers (such as

subsidies, taxes, standards)

11. The sizes of buffers and other stabilizing

stocks, relative to their flows.

10. The structure of material stocks and flows

(such as transport networks, population

age structures)

9. The lengths of delays, relative to the rate

of system change

8. The strength of negative feedback loops,

relative to the impacts they are trying to

correct against

7. The gain around driving positive feedback

loops

6. The structure of information flows (who

does and does not have access to what

kinds of information)

5. The rules of the system (such as incen-

tives, punishments, constraints)

4. The power to add, change, evolve, or self-

organize system structure


2. The mindset or paradigm out of which the

system—its goals, structure,rules, delays,

parameters—arises

1. The power to transcend paradigms


4

To explain parameters, stocks, delays,flows, feedback, and so forth, I need to startwith a basic diagram.

The “state of the system” is whateverstanding stock is of importance: amount ofwater behind the dam, amount of harvest–able wood in the forest, number of peoplein the population, amount of money in thebank, whatever. System states are usuallyphysical stocks, but they could be nonma-terial ones as well: self-confidence, degreeof trust in public officials, perceived safetyof a neighborhood.

There are usually inflows that increasethe stock and outflows that decrease it.Deposits increase the money in the bank;withdrawals decrease it. River inflow andrain raise the water behind the dam; evapo-ration and discharge through the spillwaylower it. Births and immigrations increasethe population, deaths and emigrations re-duce it. Political corruption decreases trustin public officials; experience of a well-functioning government increases it.

Insofar as this part of the system con-sists of physical stocks and flows—and theyare the bedrock of any system—it obeys lawsof conservation and accumulation. You canunderstand its dynamics readily, if you can

understand a bathtub with some water in it(the stock, the state of the system) and aninflowing faucet and outflowing drain. Ifthe inflow rate is higher than the outflowrate, the water gradually rises. If the out-flow rate is higher than the inflow, the wa-ter gradually goes down. The sluggishresponse of the water level to what could besudden twists in the input and output valvesis typical; it takes time for flows to accumu-late in stocks, just as it takes time for waterto fill up or drain out of the tub. Policychanges take time to accumulate theireffects.

The rest of the diagram shows the in-formation that causes the flows to change,which then cause the stock to change. Ifyou’re about to take a bath, you have a de-sired water level in mind (your goal). Youplug the drain, turn on the faucet, and watchuntil the water rises to your chosen level (un-til the discrepancy between the goal and theperceived state of the system is zero). Thenyou turn the water off.

If you start to get into the bath anddiscover that you’ve underestimated yourvolume and are about to produce an over-flow, you can open the drain for awhile, untilthe water goes down to your desired level.

Those are two negative feedback loops,or correcting loops, one controlling theinflow, one controlling the outflow, eitheror both of which you can use to bring thewater level to your goal. Notice that the goaland the feedback connections are not visiblein the system. If you were an extraterres-trial trying to figure out why the tub fillsand empties, it would take awhile to figureout that there’s an invisible goal and a

state ofthe system

discrepancy

inflows

perceivedstate

goal

outflows

5


discrepancy-measuring process going on inthe head of the creature manipulating thefaucets. But if you watched long enough,you could figure that out.

Very simple so far. Now let’s take intoaccount that you have two taps, a hot and acold, and that you’re also adjusting for an-other system state: temperature. Suppose thehot inflow is connected to a boiler way downin the basement, four floors below, so itdoesn’t respond quickly. And you’re mak-ing faces at yourself in the mirror and notpaying close attention to the water level. Thesystem begins to get complex, and realistic,and interesting.

Mentally change the bathtub into yourchecking account. Write checks, makedeposits, add a faucet that keeps dribbling ina little interest and a special drain that sucksyour balance even drier if it ever goes dry.Attach your account to a thousand others andlet the bank create loans as a function of yourcombined and fluctuating deposits. Link athousand of those banks into a federal re-serve system. You begin to see how simplestocks and flows, plumbed together, makeup systems way too complex to figure out.

That’s why leverage points are not in-tuitive. And that’s enough systems theoryto proceed to the list.

12. Constants, parameters,numbers

“Parameters” in systems jargon are thenumbers that determine how much of adiscrepancy turns which faucet how fast.Maybe the faucet turns hard, so it takes

awhile to get the water flowing or to turnit off. Maybe the drain is blocked and canallow only a small flow, no matter howopen it is. Maybe the faucet can deliverwith the force of a fire hose. These con-siderations are a matter of numbers, someof which are physically locked in andunchangeable, but most of which arepopular intervention points.

Consider the national debt. It’s a nega-tive bathtub, a money hole. The annual rateat which it sinks is called the deficit. Taxincome makes it rise, government expendi-tures make it fall. Congress and the Presi-dent spend most of their time arguing aboutthe many, many parameters that open andclose tax faucets and spending drains. Sincethose faucets and drains are connected tous, the voters, these are politically chargedparameters. But, despite all the fireworks,and no matter which party is in charge, themoney hole keeps getting deeper, just atdifferent rates (and even when, as in 1999,the parties are arguing about how to spenda nonexistent “surplus”).

To adjust the dirtiness of the air webreathe, the government sets parameterscalled “ambient air quality standards.” Toassure some standing stock of forest (orsome flow of money to logging companies)it sets “allowed annual cuts.” Corporationsadjust parameters such as wage rates andproduct prices, with an eye on the level intheir profit bathtub—the bottom line.

The amount of land we set aside forconservation. The minimum wage. Howmuch we spend on AIDS research or Stealthbombers. The service charge the bank ex-tracts from your account. All these are pa-


6

rameters, adjustments to faucets. So, by theway, is the act of firing people and hiringnew ones, including politicians. Putting dif-ferent hands on the faucets may change therate at which the faucets turn, but if they’rethe same old faucets, plumbed into the sameold system, turned according to the sameold information and goals and rules, thesystem isn’t going to change much. Elect-ing Bill Clinton was definitely different fromelecting Bob Dole, but not all that differ-ent, given that every President is pluggedinto the same political system.

Parameters are the points of least lever-age on my list of interventions. Diddling withthe details, arranging the deck chairs on theTitanic. Probably 90—no 95—no 99 per-cent of our attention goes to parameters, butthere’s not a lot of leverage in them.

Not that parameters aren’t important.They can be, especially in the short termand to the individual who’s standing directlyin the flow. People care deeply about pa-rameters and fight fierce battles over them.But they rarely change behavior. If the sys-tem is chronically stagnant, parameterchanges rarely kick-start it. If it’s wildly vari-able, they don’t usually stabilize it. If it’sgrowing out of control, they don’t brake it.

Whatever cap we put on campaign con-tributions, it doesn’t clean up politics. TheFeds fiddling with the interest rate haven’tmade business cycles go away. (We alwaysforget that reality during upturns, and areshocked, shocked by the downturns.) Afterdecades of the strictest air pollution stan-dards in the world, Los Angeles air is lessdirty, but it isn’t clean. Spending more onpolice doesn’t make crime go away.

Since I’m about to get into some ex-amples where parameters are leverage points,let me insert a big caveat here. Parametersbecome leverage points when they go intoranges that kick off one of the items lateron this list. Interest rates, for example, orbirth rates, control the gains around posi-tive feedback loops. System goals are param-eters that can make big differences.Sometimes a system gets onto a chaoticedge, where the tiniest change in a numbercan drive it from order to what appears tobe wild disorder.

These critical numbers are not nearly ascommon as people seem to think they are.Most systems have evolved or are designedto stay far out of critical parameter ranges.Mostly, the numbers are not worth the sweatput into them.

Here’s a story a friend sent me over theInternet to make that point:

When I became a landlord, I spent a lot of

time and energy trying to figure out what

would be a “fair” rent to charge.

I tried to consider all the variables, including

the relative incomes of my tenants, my own

income and cash flow needs, which expenses

were for upkeep and which were capital

expenses, the equity versus the interest

portion of the mortgage payments, how much

my labor on the house was worth and so on.

I got absolutely nowhere. Finally, I went to

someone who specializes in giving money

advice. She said, “You’re acting as though

there is a fine line at which the rent is fair, and

at any point above that point the tenant is

7


being screwed and at any point below that you

are being screwed. In fact, there is a large grey

area in which both you and the tenant are

getting a good, or at least a fair, deal.

Stop worrying and get on with your life.” 3

11. The sizes of buffers andother stabilizing stocks,relative to their flows

Consider a huge bathtub with slow in- andoutflows. Now think about a small one withvery fast flows. That’s the difference betweena lake and a river. You hear about catastrophicriver floods much more often than cata-strophic lake floods, because stocks that arebig, relative to their flows, are more stablethan small ones. In chemistry and other fields,a big, stabilizing stock is known as a buffer.

The stabilizing power of buffers is whyyou keep money in the bank rather thanliving from the flow of change through yourpocket. It’s why stores hold inventoryinstead of calling for new stock just ascustomers carry the old stock out the door.It’s why we need to maintain more than theminimum breeding population of an endan-gered species. Soils in the eastern U.S. aremore sensitive to acid rain than soils in thewest, because they haven’t got big buffers ofcalcium to neutralize acid.

Often you can stabilize a system byincreasing the capacity of a buffer.4 But if abuffer is too big, the system becomesinflexible. It reacts too slowly. And big buff-ers of some sorts, such as water reservoirs orinventories, cost a lot to build or maintain.

Businesses invented just-in-time invento-ries, because they figured that vulnerabilityto occasional fluctuations or screw-ups ischeaper than certain, constant inventorycosts—and because small-to-vanishinginventories allow more flexible response toshifting demand. It’s quite likely that manybusinesses making small-inventory decisionsin their own rational best interests add upto a much more unstable economy.

There’s leverage, sometimes magical, inchanging the size of buffers. But buffers areusually physical entities, not easy to change.The acid absorption capacity of eastern soilsis not a leverage point for alleviating acidrain damage. The storage capacity of a damis literally cast in concrete. So I have putbuffers at the less influential end of the listof leverage points.

10. The structure of materialstocks and flows and nodesof intersection

The plumbing structure, the stocks andflows and their physical arrangement, canhave an enormous effect on how the sys-tem operates. When the Hungarian roadsystem was laid out so all traffic from oneside of the nation to the other has to passthrough central Budapest, that determineda lot about air pollution and commutingdelays that are not easily fixed by pollutioncontrol devices, traffic lights, or speed lim-its. The only way to fix a system that is laidout wrong is to rebuild it, if you can.

Often you can’t, because physical build-ing is usually the slowest and most expen-

3 Thanks to DavidHolmstrom of Santiago,Chile.

4 For an example, seeDennis Meadows’s modelof commodity pricefluctuations: D.L.Meadows, Dynamics ofCommodity ProductionCycles. Portland, Oreg.:Productivity Press, 1970.


8

sive kind of change to make in a system.Some stock-and-flow structures are justplain unchangeable. The baby-boom swellin the U.S. population first caused pressureon the elementary school system, then highschools, then colleges, then jobs and hous-ing, and now we’re looking forward tosupporting its retirement. There is not muchwe can do about it, because five-year-oldsbecome six-year-olds, and sixty-four-year-olds become sixty-five-year-olds predictablyand unstoppably. The same can be said forthe lifetime of destructive CFC moleculesin the ozone layer, for the rate at which con-taminants get washed out of aquifers, forthe fact that an inefficient car fleet takes 10to 20 years to turn over.

Physical structure is crucial in a system,but rarely a leverage point, because chang-ing it is rarely simple. The leverage point isin proper design in the first place. After thestructure is built, the leverage is in under-standing its limitations and bottlenecks andrefraining from fluctuations or expansionsthat strain its capacity.

9. The lengths of delays,relative to the rate of systemchanges

Remember that bathtub on the fourth floorI mentioned, with the water heater in thebasement? I actually experienced one ofthose once, in an old hotel in London. Itwasn’t even a bathtub, it was a shower—nobuffering capacity. The water temperaturetook at least a minute to respond to my fau-cet twists. Guess what my shower was like.

Right, oscillations from hot to cold andback to hot, punctuated with expletives.

Delays in feedback loops are commoncauses of oscillations. If you’re trying toadjust a system state to your goal, but youonly receive delayed information about whatthe system state is, you will overshoot andundershoot. Same if your information istimely, but your response isn’t. For example,it takes several years to build an electricpower plant, and then that plant lasts, say,thirty years. Those delays make it impos-sible to build exactly the right number ofplants to supply a rapidly changing demand.Even with immense effort at forecasting,almost every centralized electricity indus-try in the world experiences long oscillationsbetween overcapacity and undercapacity. Asystem just can’t respond to short-termchanges when it has long-term delays. That’swhy a massive central-planning system, suchas the Soviet Union or General Motors,necessarily functions poorly.

Because we know they are important,we systems folks see delays wherever welook. The delay between the time when apollutant is dumped on the land and whenit trickles down to the groundwater. Thedelay between the birth of a child and thetime when that child is ready to have a child.The delay between the first successful testof a new technology and the time when thattechnology is installed throughout theeconomy. The time it takes for a price toadjust to a supply-demand imbalance.

A delay in a feedback process is criticalrelative to rates of change in the system statethat the feedback loop is trying to control.Delays that are too short cause overreaction,

9


“chasing your tail,” oscillations amplified bythe jumpiness of the response. Delays thatare too long cause damped, sustained, orexploding oscillations, depending on howmuch too long. At the extreme, they causechaos. Overlong delays in a system with athreshold, a danger point, a range past whichirreversible damage can occur, cause over-shoot and collapse.

I would list delay length as a high lever-age point, except for the fact that delays arenot often easily changeable. Things take aslong as they take. You can’t do a lot aboutthe construction time of a major piece ofcapital, or the maturation time of a child,or the growth rate of a forest. It’s usuallyeasier to slow down the change rate, so thatinevitable feedback delays won’t cause somuch trouble. That’s why growth rates arehigher up on the leverage-point list thandelay times.

And that’s why slowing economicgrowth is a greater leverage point inForrester’s world model than faster techno-logical development or freer market prices.Those are attempts to speed up the rate ofadjustment. But the world’s physical capi-tal plant, its factories and boilers, the con-crete manifestations of its workingtechnologies, can only change so fast, evenin the face of new prices or new ideas—andprices and ideas don’t change instanta-neously either, not through a whole globalculture. There’s more leverage in slowingdown the growth of the system so technolo-gies and prices can keep up with it, thanthere is in wishing the delays away.

But if there is a delay in your systemthat can be changed, changing it can have

big effects. Watch out! Be sure you changeit in the right direction! (For example, thegreat push to reduce information andmoney transfer delays in financial marketsis just asking for wild gyrations.)

8. The strength of negativefeedback loops, relative tothe impacts they are trying tocorrect against

Now we’re beginning to move from thephysical part of the system to the informa-tion and control parts, where more lever-age can be found.

Negative feedback loops are ubiquitousin systems. Nature evolves them andhumans invent them as controls to keepimportant system states within safe bounds.A thermostat loop is the classic example.Its purpose is to keep the system state called“room temperature” fairly constant at a de-sired level. Any negative feedback loop needsa goal (the thermostat setting), a monitor-ing and signaling device to detect excursionsfrom the goal (the thermostat), and aresponse mechanism (the furnace and/or airconditioner, fans, heat pipes, fuel, etc.).

A complex system usually has numerousnegative feedback loops that it can bring intoplay, so it can self-correct under different con-ditions and impacts. Some of those loops maybe inactive much of the time, like the emer-gency cooling system in a nuclear powerplant, or your ability to sweat or shiver tomaintain your body temperature. They maynot be very visible. But their presence is criti-cal to the long-term welfare of the system.


10

One of the big mistakes we make is tostrip away these “emergency” responsemechanisms because they aren’t used oftenand they appear to be costly. In the shortterm, we see no effect from doing this. Inthe long term, we drastically narrow therange of conditions over which the systemcan survive. One of the most heartbreakingways we do this is in encroaching on thehabitats of endangered species. Another isin encroaching on our own time for rest,recreation, socialization, and meditation.

The “strength” of a negative loop—itsability to keep its appointed stock at or nearits goal—depends on the combination ofall its parameters and links—the accuracyand rapidity of monitoring, the quicknessand power of response, the directness andsize of corrective flows. Sometimes there areleverage points here.

Take markets, for example, the negativefeedback systems that are all but worshippedby economists—and they can indeed bemarvels of self-correction, as prices vary tomoderate supply and demand and keepthem in balance. The more the price—thecentral piece of information signaling bothproducers and consumers—is kept clear, un-ambiguous, timely, and truthful, the moresmoothly markets will operate. Prices thatreflect full costs will tell consumers howmuch they can actually afford and will re-ward efficient producers.

Companies and governments are fatallyattracted to the price leverage point, ofcourse, all of them determinedly pushingit in the wrong direction with subsidies,fixes, externalities, taxes, and other formsof confusion.

These folks are trying to weaken thefeedback power of market signals by twist-ing information in their favor. The real le-verage here is to keep them from doing it.Hence the necessity of anti-trust laws, truth-in-advertising laws, attempts to internalizecosts (such as pollution taxes), the removalof perverse subsidies, and other ways to levelmarket playing fields.

None of which get far these days, be-cause of the weakening of another set ofnegative feedback loops: those of democ-racy. This great system was invented to putself-correcting feedback between the peopleand their government. The people, in-formed about what their elected represen-tatives do, respond by voting thoserepresentatives in or out of office. The pro-cess depends upon the free, full, unbiasedflow of information back and forth betweenelectorate and leaders. Billions of dollars arespent by leaders to limit and bias that flow.Give the people who want to distort mar-ket price signals the power to pay off thoseleaders, get the channels of communicationto be self-interested corporate partnersthemselves, and none of the necessary nega-tive feedbacks work well. Market and de-mocracy help each other erode.

The strength of a negative feedback loopis important relative to the impact it is de-signed to correct. If the impact increases instrength, the feedbacks have to be strength-ened too. A thermostat system may workfine on a cold winter day, but open all thewindows and its corrective power will fail.Democracy worked better before the adventof the brainwashing power of centralizedmass communications. Traditional controls

11


on fishing were sufficient until radar spot-ting and drift nets and other technologiesmade it possible for a few actors to wipeout the fish. The power of big industry callsfor the power of big government to hold itin check; a global economy makes neces-sary a global government.

Here are some examples of strengthen-ing negative feedback controls to improve asystem’s self-correcting abilities:

• preventive medicine, exercise, and good

nutrition to bolster the body’s ability to fight

disease;

• integrated pest management to encour-

age natural predators of crop pests;

• the Freedom of Information Act to reduce

government secrecy;

• monitoring systems to report on environ-

mental damage;

• protection of whistleblowers;

• impact fees, pollution taxes, and perfor-

mance bonds to recapture the externalized

public costs of private benefits.

7. The gain around drivingpositive feedback loops

A negative feedback loop is self-correcting;a positive feedback loop is self-reinforcing.The more it works, the more it gains powerto work some more. The more people catchthe flu, the more they infect other people.The more babies are born, the more peoplegrow up to have babies. The more moneyyou have in the bank, the more interest you

earn, the more money you have in the bank.The more the soil erodes, the less vegeta-tion it can support, the fewer roots andleaves to soften rain and run-off, the moresoil erodes. The more high-energy neutronsin the critical mass, the more they knockinto nuclei and generate more.

Positive feedback loops are sources ofgrowth, explosion, erosion, and collapse insystems. A system with an unchecked posi-tive loop ultimately will destroy itself.That’s why there are so few of them. Usu-ally a negative loop will kick in sooner orlater. The epidemic will run out ofinfectable people—or people will take in-creasingly strong steps to avoid being in-fected. The death rate will rise to equal thebirth rate—or people will see the conse-quences of unchecked population growthand have fewer babies. The soil will erodeaway to bedrock—or people will stop over-grazing, put up check dams, plant trees,and stop the erosion.

In all these examples, the first outcomeis what will happen if the positive loop runsits course, the second is what will happen ifthere is an intervention to reduce its self-multiplying power. Reducing the gainaround a positive loop—slowing thegrowth—is usually a more powerful lever-age point in systems than strengtheningnegative loops, and much preferable toletting the positive loop run.

Population and economic growth ratesare leverage points, because slowing themgives the many negative loops—technologyand markets and other forms of adaptation,all of which have limits and delays—timeto function. It’s the same as slowing the car


12

when you’re driving too fast, rather thancalling for more responsive brakes or tech-nical advances in steering.

Another example: Many positive feed-back loops in society reward the winners ofa competition with the resources to win evenbigger next time. Systems folks call them“success to the successful” loops. Rich peoplecollect interest; poor people pay it. Richpeople pay accountants and lean on politi-cians to reduce their taxes; poor people can’t.Rich people give their kids inheritances andgood educations; poor kids lose out. Anti-poverty programs are weak negative loopsthat try to counter these strong positiveones. It would be much more effective toweaken the positive loops. That’s what pro-gressive income tax, inheritance tax, anduniversal high-quality public education pro-grams are meant to do. (If rich people canbuy government and weaken, rather thanstrengthen those of measures, the govern-ment, instead of balancing “success to thesuccessful” loops, becomes just another in-strument to reinforce them!)

The most interesting behavior thatrapidly turning positive loops can trigger ischaos. This wild, unpredictable,unreplicable, and yet bounded behaviorhappens when a system starts changingmuch, much faster than its negative loopscan react to it. For example, if you keep rais-ing the capital growth rate in the worldmodel, eventually you get to a point whereone tiny increase more will shift theeconomy from exponential growth to os-cillation. Another nudge upward gives theoscillation a double beat. And just the tiniestfurther nudge sends it into chaos.

I don’t expect the world economy toturn chaotic any time soon (not for thatreason, anyway). That behavior occurs onlyin unrealistic parameter ranges, equivalentto doubling the size of the economy withina year. Real-world systems can turn chaotic,however, if something in them can grow ordecline very fast. Fast-replicating bacteria orinsect populations, very infectious epidem-ics, wild speculative bubbles in money sys-tems, neutron fluxes in the guts of nuclearpower plants; these systems can turn cha-otic. Control must involve slowing downthe positive feedbacks.

In more ordinary systems, look for le-verage points around birth rates, interestrates, erosion rates, “success to the success-ful” loops, any place where the more youhave of something, the more you have thepossibility of having more.

6. The structure ofinformation flows

There was this subdivision of identicalhouses, the story goes, except that for somereason the electric meter in some of thehouses was installed in the basement and inothers it was installed in the front hall, wherethe residents could see it constantly, goinground faster or slower as they used more orless electricity. With no other change, withidentical prices, electricity consumption was30 percent lower in the houses where themeter was in the front hall.

We systems-heads love that story be-cause it’s an example of a high leverage pointin the information structure of the system.

13


It’s not a parameter adjustment, not astrengthening or weakening of an existingloop. It’s a new loop, delivering informationto a place where it wasn’t going before andtherefore causing people to behavedifferently.

A more recent example is the ToxicRelease Inventory, the U.S. government’srequirement, instituted in 1986, that everyfactory releasing hazardous air pollutantsreport those emissions publicly every year.Suddenly every community could find outprecisely what was coming out of the smoke-stacks in town. There was no law againstthose emissions, no fines, no determinationof “safe” levels, just information. But by1990, emissions dropped 40 percent. Theyhave continued to go down since, not somuch because of citizen outrage as becauseof corporate shame. One chemical companythat found itself on the Top Ten Polluterslist reduced its emissions by 90 percent, justto “get off that list.”

Missing feedback is one of the mostcommon causes of system malfunction.Adding or restoring information can be apowerful intervention, usually much easierand cheaper than rebuilding physicalinfrastructure. The tragedy of the commonsthat is crashing the world’s commercial fish-eries occurs because there is no feedbackfrom the state of the fish population to thedecision to invest in fishing vessels. (Con-trary to economic opinion, the price of fishdoesn’t provide that feedback. As the fishget more scarce and hence more expensive,it becomes all the more profitable to go outand catch them. That’s a perverse feedback,a positive loop that leads to collapse.)

It’s important that the missing feedbackbe restored to the right place and in compel-ling form. To take another tragedy of thecommons, it’s not enough to inform all theusers of an aquifer that the groundwater levelis dropping. That could initiate a race to thebottom. It would be more effective to set awater price that rises steeply as the pumpingrate begins to exceed the recharge rate.

Compelling feedback. Suppose taxpay-ers could specify on their return forms whatgovernment services their tax paymentsmust be spent on. (Radical democracy!)Suppose any town or company that puts awater intake pipe in a river had to put itimmediately downstream from its own out-flow pipe. Suppose any public or private of-ficial who made the decision to invest in anuclear power plant got the waste from thatplant stored on his/her lawn. Suppose (thisis an old one) that the politicians who de-clare war were required to spend that warin the front lines.

We humans have a systematic tendencyto avoid accountability for our own deci-sions. That’s why so many feedback loopsare missing—and why this kind of leveragepoint is so often popular with the masses,unpopular with the powers that be, and ef-fective, if you can get the powers that be topermit it to happen (or go around them andmake it happen anyway).

5. The rules of the system

The rules of the system define its scope, itsboundaries, its degrees of freedom. Thoushalt not kill. Everyone has the right of free


14

speech. Contracts are to be honored. ThePresident serves four-year terms and can-not serve more than two of them. Ninepeople on a team, you have to touch everybase, three strikes and you’re out. If you getcaught robbing a bank, you go to jail.

Mikhail Gorbachev came to power inthe USSR, opened information flows(glasnost), changed the economic rules(perestroika), and look what happened.

Constitutions are strong social rules.Physical laws such as the second law ofthermodynamics are absolute rules, whetherwe understand them or not, or like them ornot. Laws, punishments, incentives, andinformal social agreements are progressivelyweaker rules.

To demonstrate the power of rules, I liketo ask my students to imagine different onesfor a college. Suppose the students gradedthe teachers, or each other. Suppose therewere no degrees: you come to college whenyou want to learn something, and you leavewhen you have learned it. Suppose tenurewere awarded to professors according totheir ability to solve real-world problems,rather than publishing academic papers.Suppose a class was graded as a group,instead of as individuals.

As we try to imagine restructured ruleslike these and what our behavior would beunder them, we come to understand thepower of rules. They are high leveragepoints. Power over the rules is real power.That’s why lobbyists congregate when Con-gress writes laws, and why the SupremeCourt, which interprets and delineates theConstitution—the rules for writing therules—has even more power than Congress.

If you want to understand the deepest mal-functions of systems, pay attention to therules, and to who has power over them.

That’s why my system intuition was send-ing off alarm bells while the new world tradesystem was explained to me. It is a systemwith rules designed by corporations, run bycorporations, for the benefit of corporations.Its rules exclude almost any feedback fromany other sector of society. Most of its meet-ings are closed even to the press (no infor-mation flow, no feedback). It forces nationsinto positive loops “racing to the bottom,”competing with each other to weaken envi-ronmental and social safeguards in order toattract investment and trade. It’s a recipe forunleashing “success to the successful” loops,until they generate enormous accumulationsof power and huge centralized planning sys-tems that will destroy themselves, just as theSoviet Union destroyed itself, and for simi-lar systemic reasons.

4. The power to add, change,evolve, or self-organizesystem structure

The most stunning thing living systems andsocial systems can do is to change themselvesutterly by creating whole new structures andbehaviors. In biological systems that poweris called evolution. In human society it’scalled technical advance or social revolution.In systems lingo, it’s called self-organization.

Self-organization means changing anyaspect of a system lower on this list: addingcompletely new physical structures, such asbrains or wings or computers; adding new

15


negative or positive loops: making new rules.The ability to self-organize is the strongestform of system resilience. A system that canevolve can survive almost any change, bychanging itself. The human immune sys-tem has the power to develop new responsesto (some kinds of ) insults it has never be-fore encountered. The human brain can takein new information and pop out completelynew thoughts.

The power of self-organization seems sowondrous that we tend to regard it as mys-terious, miraculous, manna from heaven.Economists often model technology as lit-eral manna, coming from nowhere, costingnothing, increasing the productivity of aneconomy by some steady percent each year.For centuries, people have regarded the spec-tacular variety of nature with the same awe.Only a divine creator could bring forth sucha creation.

Further investigation of self-organizingsystems reveals that the divine creator, ifthere is one, did not have to produce evolu-tionary miracles. He, she, or it just had towrite marvelously clever rules for self-organization. These rules basically governhow, where, and what the system can addonto or subtract from itself under what con-ditions. As hundreds of self-organizing com-puter models have demonstrated, complexand delightful patterns can evolve fromquite simple evolutionary algorithms. (Thatneed not mean that real-world algorithmsare simple, only that they can be.) Thegenetic code within the DNA that is thebasis of all biological evolution contains justfour different “letters”, combined into“words” of three letters each. That pattern,

and the rules for replicating and rearrang-ing it, has been constant for something likethree billion years, during which it hasspewed out an unimaginable variety of failedand successful self-evolved creatures.

Self-organization is basically the combi-nation of an evolutionary raw material—ahighly variable stock of information fromwhich to select possible patterns—and ameans for experimentation, for selecting andtesting new patterns. For biological evolu-tion the raw material is DNA, one source ofvariety is spontaneous mutation, and the test-ing mechanism is something like punctuatedDarwinian selection. For technology, the rawmaterial is the body of understanding peoplehave accumulated and stored in libraries andin brains. The source of variety is human cre-ativity (whatever that is) and the selectionmechanism can be whatever the market willreward or whatever governments and foun-dations will fund or whatever meets humanneeds or solves an immediate problem.

When you understand the power ofsystem self-organization, you begin to un-derstand why biologists worship biodiversityeven more than economists worship tech-nology. The wildly varied stock of DNA,evolved and accumulated over billions ofyears, is the source of evolutionary poten-tial, just as science libraries and labs anduniversities where scientists are trained arethe source of technological potential. Allow-ing species to go extinct is a systems crime,just as randomly eliminating all copies ofparticular science journals, or particularkinds of scientists, would be.

The same could be said of humancultures, of course, which are the store of


16

behavioral repertoires, accumulated overnot billions, but hundreds of thousands ofyears. They are a stock out of which socialevolution can arise. Unfortunately, peopleappreciate the precious evolutionary poten-tial of cultures even less than they under-stand the preciousness of every geneticvariation in the world’s ground squirrels. Iguess that’s because one aspect of almostevery culture is the belief in the uttersuperiority of that culture.

Insistence on a single culture shuts downlearning. Cuts back resilience. Any system,biological, economic, or social, that becomesso encrusted that it cannot self-evolve, asystem that systematically scorns experimen-tation and wipes out the raw material ofinnovation, is doomed over the long termon this highly variable planet.


The goal of a system is a leverage point su-perior to the self-organizing ability of a sys-tem. For example, if the goal is to bring moreand more of the world under the control ofone particular central planning system (theempire of Genghis Khan, the world of Is-lam, the People’s Republic of China, Wal-Mart, Disney, whatever), then everythingfurther down the list, physical stocks andflows, feedback loops, information flows,even self-organizing behavior, will be twistedto conform to that goal.

That’s why I can’t get into argumentsabout whether genetic engineering is a “good”or a “bad” thing. Like all technologies, it de-pends upon who is wielding it, with what

goal. The only thing one can say is that ifcorporations wield it for the purpose of gen-erating marketable products, that is a verydifferent goal, a different selection mecha-nism, a different direction for evolution thananything the planet has seen so far.

As my little single-loop examples haveshown, most negative feedback loops withinsystems have their own goals: to keep thebathwater at the right level, to keep theroom temperature comfortable, to keepinventories stocked at sufficient levels, tokeep enough water behind the dam. Thosegoals are important leverage points for piecesof systems, and most people realize that. Ifyou want the room warmer, you know thethermostat setting is the place to intervene.But there are larger, less obvious, higher-leverage goals, those of the entire system.

Whole system goals are not what wethink of as goals in the human-motivationalsense. They are not so much deducible fromwhat anyone says as from what the systemdoes. Survival, resilience, differentiation,evolution are system-level goals.

Even people within systems don’t oftenrecognize what whole-system goal they areserving. To make profits, most corporationswould say, but that’s just a rule, a necessarycondition to stay in the game. What is thepoint of the game?

To increase stockholder wealth, mosteveryone would say, and that is a powerful,behavior-shaping goal. But there is an evenlarger one, formally espoused by no one, butobvious when one looks at the actual be-havior of the system. To grow, to increasemarket share, to bring the world (custom-ers, suppliers, regulators) more and more

17


under the control of the corporation, so thatits operations become ever more shieldedfrom uncertainty. John Kenneth Galbraithrecognized that corporate goal—to engulfeverything —long ago.5

It’s the goal of a cancer cell too. Actu-ally it’s the goal of every living population,and only a bad one when it isn’t balancedby higher-level negative feedback loops thatnever let an upstart power-driven entitycontrol the world. The goal of keeping themarket competitive has to trump the goalof each corporation to eliminate its com-petitors (and brainwash its customers andswallow its suppliers), just as in ecosystems,the goal of keeping populations in balanceand evolving has to trump the goal of eachpopulation to reproduce without limit andcontrol all the resource base.

I said earlier that changing the playersin the system is a low-level intervention, aslong as the players fit into the same old sys-tem. The exception to that rule is at the top,where a single player can have the power tochange the system’s goal. I have watched inwonder as—only very occasionally—a newleader in an organization, from DartmouthCollege to Nazi Germany, comes in, enun-ciates a new goal, and swings hundreds orthousands or millions of perfectly intelli-gent, rational people off in a new direction.

That’s what Ronald Reagan did. Notlong before he came to office, a Presidentcould say “Ask not what government cando for you, ask what you can do for thegovernment,” and no one even laughed.Reagan said over and over, the goal is not toget the people to help the government andnot to get government to help the people,

but to get government off our backs. Onecan argue, and I would, that larger systemchanges and the rise of corporate power overgovernment let him get away with that. Butthe thoroughness with which the public dis-course in the U.S. and even the world hasbeen changed since Reagan is testimony tothe high leverage of articulating, meaning,repeating, standing up for, insisting, forbetter or for worse, upon new system goals.

2. The mindset or paradigmout of which the systemarises

Another of Jay Forrester’s famous systemssayings goes: It doesn’t matter how the taxlaw of a country is written. There is a sharedidea in the minds of the society about whata “fair” distribution of the tax load is. What-ever the rules say, by fair means or foul, bycomplications, cheating, exemptions or de-ductions, by constant sniping at the rules,actual tax payments will push right upagainst the accepted idea of “fairness.”

The shared idea in the minds of society,the great big unstated assumptions—un-stated because unnecessary to state; every-one already knows them—constitute thatsociety’s paradigm, or deepest set of beliefsabout how the world works.

There is a difference between nouns andverbs. Money measures something real andhas real meaning (therefore people who arepaid less are literally worth less). Growth isgood. Nature is a stock of resources to beconverted to human purposes. Evolutionstopped with the emergence of Homo

5 John KennethGalbraith, The NewIndustrial State. Boston:Houghton Mifflin,1967.


18

sapiens. One can “own” land. Those are justa few of the paradigmatic assumptions ofour current culture, all of which have ut-terly dumfounded other cultures, whothought them not the least bit obvious.

Paradigms are the sources of systems.From them, from shared social agreementsabout the nature of reality, come systemgoals and information flows, feedbacks,stocks, flows and everything else aboutsystems. No one has ever said that betterthan Ralph Waldo Emerson:

Every nation and every man instantly

surround themselves with a material

apparatus which exactly corresponds to . . .

their state of thought. Observe how every

truth and every error, each a thought of some

man’s mind, clothes itself with societies,

houses, cities, language, ceremonies,

newspapers. Observe the ideas of the present

day . . . see how timber, brick, lime, and

stone have flown into convenient shape,

obedient to the master idea reigning in the

minds of many persons. . . . It follows, of

course, that the least enlargement of ideas . . .

would cause the most striking changes of

external things.6

The ancient Egyptians built pyramids be-cause they believed in an afterlife. We buildskyscrapers because we believe that space indowntown cities is enormously valuable.(Except for blighted spaces, often near theskyscrapers, which we believe are worthless.)Whether it was Copernicus and Keplershowing that the earth is not the center ofthe universe, or Einstein hypothesizing thatmatter and energy are interchangeable, or

Adam Smith postulating that the selfish ac-tions of individual players in markets won-derfully accumulate to the common good,people who have managed to intervene insystems at the level of paradigm have hit aleverage point that totally transforms systems.

You could say paradigms are harder tochange than anything else about a system,and therefore this item should be lowest onthe list, not second-to-highest. But there’snothing necessarily physical or expensive oreven slow in the process of paradigm change.In a single individual it can happen in amillisecond. All it takes is a click in themind, a falling of scales from eyes, a newway of seeing. Whole societies are anothermatter. They resist challenges to their para-digm harder than they resist anything else.Societal responses to paradigm challenge haveincluded crucifixions, burnings at the stake,concentration camps, and nuclear arsenals.

So how do you change paradigms? Tho-mas Kuhn, who wrote the seminal bookabout the great paradigm shifts of science,has a lot to say about that. 7 In a nutshell,you keep pointing at the anomalies and fail-ures in the old paradigm, you keep speak-ing louder and with assurance from the newone, you insert people with the new para-digm in places of public visibility and power.You don’t waste time with reactionaries;rather you work with active change agentsand with the vast middle ground of peoplewho are open-minded.

Systems folks would say you change para-digms by modeling a system on a computer,which takes you outside the system and forcesyou to see it whole. We say that because ourown paradigms have been changed that way.

6 Ralph Waldo Emerson,“War” (lecture delivered inBoston, March 1838).Reprinted in Emerson’sComplete Works, vol. XI.Boston: Houghton,Mifflin & Co., 1887, p.177.

7 Thomas Kuhn, TheStructure of ScientificRevolution. Chicago:University of ChicagoPress, 1962.

19


1. The power to transcendparadigms

There is yet one leverage point that is evenhigher than changing a paradigm. That isto keep oneself unattached in the arena ofparadigms, to stay flexible, to realize thatno paradigm is “true,” that every one,including the one that sweetly shapes yourown worldview, is a tremendously limitedunderstanding of an immense and amaz-ing universe that is far beyond human com-prehension. It is to “get” at a gut level theparadigm that there are paradigms, and tosee that that itself is a paradigm, and toregard that whole realization asdevastatingly funny. It is to let go into NotKnowing, into what the Buddhists callenlightenment.

People who cling to paradigms (justabout all of us) take one look at the spa-cious possibility that everything they thinkis guaranteed to be nonsense and pedal rap-idly in the opposite direction. Surely thereis no power, no control, no understand-ing, not even a reason for being, much lessacting, in the notion or experience thatthere is no certainty in any worldview. But,in fact, everyone who has managed to en-tertain that idea, for a moment or for alifetime, has found it to be the basis forradical empowerment. If no paradigm isright, you can choose whatever one willhelp to achieve your purpose. If you haveno idea where to get a purpose, you canlisten to the universe (or put in the nameof your favorite deity here) and do his, her,its will, which is probably a lot betterinformed than your will.

It is in this space of mastery over para-digms that people throw off addictions, livein constant joy, bring down empires, foundreligions, get locked up or “disappeared”or shot, and have impacts that last formillennia.

A final caution

Back from the sublime to the ridiculous, fromenlightenment to caveats. So much has to besaid to qualify this list. It is tentative and itsorder is slithery. Every item has exceptionsthat can move it up or down the order ofleverage. Having had the list percolating inmy subconscious for years has not trans-formed me into a Superwoman. The higherthe leverage point, the more the system willresist changing it—that’s why societies tendto rub out truly enlightened beings.

Magical leverage points are not easilyaccessible, even if we know where they areand which direction to push on them. Thereare no cheap tickets to mastery. You have towork at it, whether that means rigorouslyanalyzing a system or rigorously casting offyour own paradigms and throwing yourselfinto the humility of Not Knowing. In theend, it seems that power has less to do withpushing leverage points than it does withstrategically, profoundly, madly letting go.

places to intervene in a system - the donella meadows ... · 1 folks who do systems analysis have a...

Documents