Skip to content

Skeptics and thermostats

UPDATE 12:50PM: Please assume I’m an idiot (see end of post)

Many have suffered from being in a building where there was a centralized thermostat for the whole building (or the whole floor), with the predictable result that some rooms are way too hot or way too cold. (Sounds like a metaphor, watch for it…)

Things were even more extreme in the former Soviet Union, where there were centralized heating plants for a whole city, and the hot air would then be pumped out to individual homes and offices. So basically the whole city had one centralized thermostat.

Krasnoyarsk thermal power station Number 1, in Siberia. In 2008 during winter temperatures of -4 degrees, a burst pipeline at this station cut off central heat for days for some 10,000 people. Source: REUTERS/Ilya Naymushin (RUSSIA)

What a nice and simple solution there is: give each room its own thermostat. First, there is automatic adjustment from the thermostat to keep it from being too hot or too cold. Second, the people in the room at any one moment can choose to adjust the thermostat according to their preferences.

A thermostat is a very simple knowledge processing device. So this is a great metaphor for (here it comes!) the advantages of decentralized knowledge over centralized knowledge  (Hat tip to Adam Martin for the Facebook conversation that sparked this idea).

When skeptics (like me) criticize the uselessness of very aggregated centralized knowledge on “how to do development”, we get labeled nihilists, like we’re saying nobody never knows nothing nowhere nohow. But what we’re really saying is that centralized knowledge is an impossible dream for overall economic development, but decentralized knowledge can work very well.

In sum:

1) Skeptics like me are not criticizing ALL knowledge, just saying some types are useful, and others are not. And so the best systems are those that can gather and process decentralized knowledge.

2) Well-functioning markets and democracy give people their own thermostats.

PS {Insert here your own favorite example of the centralized approach to global problems at Davos starting today.}

UPDATE 12:50pm on Please assume an idiot:

In response to commentators:

(1) have I ever heard of any situation where centralized knowledge plays a full or partial role? Yes

(2) does that change the above argument? No

This entry was posted in Big ideas, Economics principles. Bookmark the permalink. Follow any comments here with the RSS feed for this post. Both comments and trackbacks are currently closed.


  1. jeff barnes wrote:

    Centralized knowledge?? Isnt the metaphor really about centralized vs. decentralized decision making? Whether to let one person to set the thermostat for everyone or let everyone decide for themselves. It seems to me as an academic you would recognize that knowledge has to be centralized before it can be disseminated.

    Posted January 26, 2011 at 7:29 am | Permalink
  2. lazyfair wrote:

    Jeff, why must knowledge be centralized?

    Posted January 26, 2011 at 10:33 am | Permalink
  3. Roger wrote:

    Decentralising knowledge means fragmenting it not only locally, geographically but also at the individual level to a point where it becomes meaningless. So yes centralised knowledge may be useless at the lcoal level but decentralised knowledge can be inefective at the global level.

    Look at HIV, UNAIDS, the 3 “one”, then at each and every little subgroup that claim and require attention that neither central nor local knwoledge can answer.

    Tehir is a need for a shuttle between people and knowledgeS in order for the system to work and maximise output. Not sure we have found one yet!

    Posted January 26, 2011 at 10:33 am | Permalink
  4. Steve S wrote:


    The knowledge of the exact temperature of each room in the city can only be known by the individual inhabitants of each room. Yes, decision making is tied into this, but assuming that “it’s cold outside, crank up the heat” is a good use of centralized knowledge fails to take into account myriad factors.

    One room may have people who like it at 60. One room may like it at 75. One room may rather wear sweaters than pay to keep the room at 70.

    The centralized “it’s cold” leads to inefficient use of resources and more unhappy people.

    Decisions are predicated on knowledge, so you can’t dissociate the two as you did.

    Posted January 26, 2011 at 11:25 am | Permalink
  5. Steve S wrote:

    Also forgot to mention that not only the preferences of absolute temperature matter, but the conditions of each building requires decentralized knowledge as well.

    If you are an apartment in the middle of the building you don’t need as much heat as one on the exterior, or one with a lot of windows, etc.

    Posted January 26, 2011 at 11:28 am | Permalink
  6. David Week wrote:

    Roger writes “Decentralising knowledge means fragmenting it not only locally, geographically but also at the individual level to a point where it becomes meaningless.” That’s just plain silly. I know my wife, and my children, at an individual level. That knowledge is central to my life. It is not “meaningless”. Also, I know how to write about my work in a way that no-one else on the planet does. Again: not meaningless.

    The whole debate is best understood in terms of “universal knowledge”, which was invented by the Greeks, and local knowledge, the subject of a book of that title by Clifford Geertz, q.v. This distinction is also embodied in the awful-sounding postmodern neologism “knowledges”.

    Both universal knowledge and local knowledge have domains and situations in which each is useful. Universal knowledge is useful in bridge building (though even there, a bit of local knowledge is absolutely necessary.) Local knowledge is useful in wooing a spouse (though “pickup artists” will try and tell you that they have universal knowledge to that effect.)

    Local knowledge is never “useless”. It’s only ever useful, though, to the people that hold it, which are “local” people. Development requires local knowledge. That forces development funders to consider delegating expenditure authority to the field, and even to the beneficiaries. They don’t like this. Therefore, there is an enormous institutional thrust to generate universal knowledge for development. This thrust comes from the interest of funders, who want it for their own comfort and control, not the beneficiaries, who usually have a pretty good idea of what needs to be done, being intimately familiar with the condition of their own lives.

    Posted January 26, 2011 at 11:31 am | Permalink
  7. Tom of the Missouri wrote:


    The shuttle is called the internet. Before that it was the telephone, the telegraph, the pony express, the silk road, word of mouth, etc. Notice a pattern here? Hint: with each iteration societies become more wealthy. Another hint: The info does not just flow from central authorities to us dumb decentralized little people. It now flows in all directions24/7, including to and from your small African viliages with HIV, where the inhabitants can now better make their own decisions. You may continue taking marching orders from your preferred central authority, but I think I will continue to get and share my info from a million decentralized sources. All of these “shuttles” as you call them allow more decision to be made in a decentralized fashion. The thermostadt analogy is very apt. btw.

    Posted January 26, 2011 at 11:50 am | Permalink
  8. PS wrote:

    Actually, centralized heating requires both. Without centralized knowledge about individual decisions, how can the building owner/manager know whether the pipeline to the centralized heating plant is sufficient? The heating plant also represents a centralized knowledge that can anticipate overall demand based on the weather, i.e., knows to crank up the plant as winter approaches.

    Posted January 26, 2011 at 12:31 pm | Permalink
  9. David Week wrote:

    Eugene Smith, Albert Schweitzer, moral dilemmas, local knowledge:

    Posted January 26, 2011 at 12:46 pm | Permalink
  10. Roger wrote:

    Currently, central knowledge is built by piecing together bits of local knowledge collected here and there by different people in different circumstances using different tools. From there, this central knowledge which has more in common with dogma than knowledge is then used locally to determine the nature of an intervention.

    A typical example using HIV prevention is that HIV disproportionately affects women (dogma, until recently) therefore let’s design and implement HIV prevention programme targeting in priority women. Well, that’s unfortunately not true. HIV does indeed disproportionally affect women … in sub Saharan Africa but elsewhere, it is men (particularly those who have sex with other men) or IDU who are mostly affected (former Soviet Union for example, but also Vietnam as another example).

    That’s where the local knowledge comes into the picture and where one realises that, still with HIV prevention mind (that’s what I am interested in), you can’t put people in the same box and that if you want to be really efficient then you need to have knowledge on an individual basis. Not all MSM or IDU are vulnerable or at the same risk of HIV infection. This is why I wrote that fragmented at the individual level knowledge becomes meaningless (not useless!) because it does not lead to clear-cut intervention that can target more than one individual, because not only people can’t be put in silo, but even if it was possible they tend to freely move from one silo to another. But fragemented knowledge is ceratinly not useless for understanding that programme centrally design based on dogma will certainly fail.

    I also certainly don’t disagree with the thermostat analogy, and on decentralising action and developemnt intervention, on the contrary, but if two people in the same room have a different perception of how hot or cold they feel, how having a thermostat in the room help? Who has priority on opening or closing the window?

    There is nothing new in wanting to decentralise development, after all isn’t it was PRA was about? Now there is a lot of litterature available about the success and failure of PRA and we coudl probably avoid repeating past mistake under a new name.

    As for the role and value of the internet as a shuttle for HIV prevention in an African village, I don’t think we are there yet, but that’s another debate.

    Posted January 26, 2011 at 12:50 pm | Permalink
  11. Chops wrote:

    Centralized thermostats at my economics department building lead to many offices having open windows during the winter to keep the heat down. Waste energy much?

    Posted January 26, 2011 at 4:51 pm | Permalink
  12. todd wrote:

    While not a proponent of a single thermostat, I find the shortcoming in your scenario and the analogy you are trying to construct is that the idea that the temperature of one room has no affect on the occupants of rooms adjacent is not true. If one room wants piping hot this has some impact on the room next door. It may mean that the next room can turn its heat to much lower than it would otherwise need or it may mean it needs to turn on cool air and this may result in a spiraling feedback loop of each room using way more energy than it needs. Would it not make sense for there to be some type of global management to ensure that damaging types of interactions don’t occur?

    Posted January 27, 2011 at 12:06 am | Permalink
  13. Kabir wrote:

    I’m glad Adam on facebook gets a hat-tip, but what about Hayek, the pioneer in work on the importance of decentralized knowledge and decision making?!

    I’m surprised anyone would question the merits of decentralized knowledge and decision making. I feel like the more important question becomes the costs associated with such a system. While Chops’ econ department may be wasting energy, what is the cost of installing individual control in each room? (I’m not saying it isn’t worth it in this example…probably is. But just trying to understand when it is worthwhile to actually make a shift from one system to another)

    Posted January 27, 2011 at 2:36 am | Permalink
  14. Paul C wrote:

    Another problem with the metaphor is that district heating systems are more effective, more efficient and more environmentally friendly at providing heat at scale. Poor implementation in former communist countries doesn’t mean that the technology doesn’t work well.

    Just because your metaphor is wrong doesn’t mean your point is wrong; however it does illustrate the importance of knowledge before judgement. Unfortunately in this case, the utility of district heating does appear to undermine your essential point &c &c.

    I like your blog anyway 😀

    Posted January 27, 2011 at 6:01 am | Permalink
  15. James B wrote:

    Is the debate here about local knowledge vs. central memory as opposed to central planning? Of course the central memory is required for there to be the retained idea of the thermostat with the stuff at the other end conrolled by the thermostat, but the local knowledge produces the ideal conditions using this tool produced from collective memory. It is a good analogy. By the way, who does not despise being in a room where the temperature cannot be controlled for conditions at the moment? This is only resentment at having others decide from a distance what we must endure.

    District heating is efficient? Compared to what? Producing steam in hopes someone will take it up is not particularly modern. It is difficult to keep steam in that state – meaning many calories or BTUs if you will must be used to do so. If the steam is waste steam, fine; but why not use the waste steam to instead turn a heat recovery steam generator and turbine to produce electricty that can be used to produce among many other things? Also, for what it’s worth, the steam is ordinarily made from the calories or BTUs in coal.

    Posted January 27, 2011 at 7:42 am | Permalink
  16. Paul C wrote:

    “District heating is efficient? Compared to what?”

    Compared to the alternatives. I am of course talking about district heating that uses cogeneration, which is the model most often used today.

    Posted January 27, 2011 at 9:02 am | Permalink
  17. ralph eggen wrote:

    I have a small, 8 units, apartment building with central thermostat. Each apartment is equipped with radiators each of which has an adjustable valve to regulate the heat desired. The central thermostat establishes the top limit for heat. If I see an open window, I talk with the tenant to point out the use of the valve on each radiator.
    Works for me.

    Posted January 27, 2011 at 10:03 am | Permalink
  18. Johnny D wrote:

    Todd wrote “Would it not make sense for there to be some type of global management to ensure that damaging types of interactions don’t occur?”

    How would global management even be aware of the interaction? Just the opposite, the global management adds one more control loop into the system which further destabilizes it. It’s one more degree of freedom in the system.

    Posted January 27, 2011 at 7:38 pm | Permalink
  19. Raphael wrote:

    As Roger notes, decentralized knowledge is
    the whole point of Participatory Rural Appraisal (PRA). The irony is that the industry has now moved towards quantitative methods prized by economists (methods which tend to be geared towards central decisionmakers looking for external validity), rather than the locally rich info we get from PRA. So, in sum, economists are the perpetuators of centralized knowledge. 😉

    Posted January 28, 2011 at 4:25 pm | Permalink
  20. David Dillon wrote:

    This whole business about universal versus
    decentralized knowledge misses the point. The great advantage of a market civilizaion is that it allows the pooling of knowledge possessed by many people to accomplish what no one of them could accomplish alone.
    The fable “I:Pencil” shows this quite graphically.

    Posted January 30, 2011 at 1:27 pm | Permalink

6 Trackbacks

  1. […] This post was mentioned on Twitter by William Easterly, Steve Kowal and OpenEye Group, Conduit Journal. Conduit Journal said: Skeptics and thermostats […]

  2. By Skeptics and thermostats | Global Health Hub on January 26, 2011 at 5:22 am

    […] the rest here: Skeptics and thermostats AKPC_IDS += "10489,";Popularity: 50% […]

  3. By Skeptics and thermostats « Daniel Smith on January 26, 2011 at 9:35 am
  4. By Some Links on January 26, 2011 at 10:01 am

    […] Bill Easterly uses a truly inspired analogy to explain the importance of Hayek’s insight about… (and, I would add, also of the institution of private property). […]

  5. […] comment was originally posted here. blog comments powered by Disqus /* You can follow me on twitter here: […]

  6. By The Daily Eudemon on January 31, 2011 at 1:23 am

    […] and don’t apologize. * * * * * * * Hayek Corner. The superiority of decentralized knowledge. An analogy, using the thermostat. * * * * * * * Cut ‘Em Up. Credit card “interest rates are now hovering near record […]

  • About Aid Watch

    The Aid Watch blog is a project of New York University's Development Research Institute (DRI). This blog is principally written by William Easterly, author of "The Elusive Quest for Growth: Economists' Adventures and Misadventures in the Tropics" and "The White Man's Burden: Why the West's Efforts to Aid the Rest Have Done So Much Ill and So Little Good," and Professor of Economics at NYU. It is co-written by Laura Freschi and by occasional guest bloggers. Our work is based on the idea that more aid will reach the poor the more people are watching aid.

    "Conscience is the inner voice that warns us somebody may be looking." - H.L. Mencken

  • Archives