La Bourse / Volume VI / Nº VI.02 / Workbook — the student

A woman standing at the far end of a glass boardroom table, speaking to the board seated on either side. — Plate VI.02 · Workbook — the studentThe Thursday Tribunal.The Tribunal de las Aguas has met outside the cathedral in Valencia, on Thursdays, for roughly a thousand years. It keeps no written record of its judgments. It has never needed one, because everybody who matters was standing there.

WORKBOOK — THE STUDENT

Chapter VI.02 · Ostrom, Fully Applied

For the person studying this alone, or in a seminar, with no organisation to govern yet. You have something better than an organisation: you have time, and you have access to more primary data about real commons than any researcher had before about 2005. This workbook puts both to work.

WHY THIS WORKBOOK IS DIFFERENT

Most students meet the eight design principles as a slide and leave with a slide. This workbook has you do the two things that turn the slide into an instrument you own: read the original wording and notice what the paraphrase dropped, and score a real commons with numbers you fetched yourself.

The second is the important one and it is genuinely available to you. Every digital commons on earth publishes its own census. You can read English Wikipedia's entire governance apparatus — every policy, every sanction, every arbitration decision, every edit ever made — without asking anyone's permission, from a laptop, tonight. There is no equivalent access to a fishery.

So you are not studying a smaller version of what a researcher does. For this particular topic, in this particular decade, you have the same access they have, and the only thing between you and a publishable finding is that nobody has done the counting.

PART ONE — DISCOVERY

Weeks one to four: read the original, then find the commons around you

Exercise 1.1 — The wording audit (90 minutes)

Find the eight principles as Ostrom wrote them — Table 3.1, page 90 of Governing the Commons. Then find three paraphrases in the wild: a lecture slide, a consultancy blog, a Wikipedia summary.

For each of the eight, write two columns: what Ostrom wrote, and what the paraphrase says. Then mark every qualifier the paraphrase dropped.

You are looking for three in particular, and you should find at least two:

Principle 4 is a compound claim — audit resource condition and user behaviour, and be accountable to the audited. Most paraphrases reduce it to "have monitoring."
Principle 5's graduation is by seriousness and context of the offence, not by repeat count.
Principle 7 is a negative — external authorities do not challenge the right. Most paraphrases upgrade it to "recognition" or "support", which is a much stronger and much rarer condition.

Output: a one-page table. Keep it. You will use it every time somebody cites the principles at you for the rest of your career.

Exercise 1.2 — Five commons within a mile of you (one week)

Not metaphorical ones. Find five actual shared resources where appropriation by one party reduces what is available to another, or where provision is under-supplied relative to appropriation.

Candidates that are almost always present: a shared kitchen or fridge; a laundry room; a bicycle store; a departmental printing budget; a group project's shared document; a lab's equipment booking system; a sports club's kit; a communal garden; a shared car; a course's reading-list copies in the library.

For each, write four lines: what the resource is, who the appropriators are, who provides, and what happens when somebody takes more than their share.

Do not evaluate yet. You are looking for five so that you can choose one.

Exercise 1.3 — The unwritten rule (60 minutes)

Take the one of the five that works best, and find the rule that nobody wrote down. There will be one. It will usually be a proxy — something visible that stands in for something invisible.

Törbel's is the purest in the literature: no villager may send more cows to the summer pasture than they can overwinter on hay from their own land. Nobody audits the pasture. Everybody can see the barn.

Write your commons' hay rule. If you cannot find one, that is the finding, and it will predict which of the twelve that commons scores badly on.

PART TWO — THE ARITHMETIC

Weeks five to eight: score something real

Exercise 2.1 — Fetch a census (2 hours)

Pick a digital commons with a public API or a public statistics page. The English-language Wikipedia is the richest, and one request returns the whole census:

  action=query & meta=siteinfo & siprop=statistics & format=json

The chapter's figures came from exactly that request on 17 September 2026:

  articles                              7,240,827
  pages, all namespaces                66,269,160
  edits, lifetime                   1,370,633,170
  registered accounts                  54,539,294
  active accounts (30 days)               262,745
  administrators                              809

Run it yourself and get different numbers. They move by the minute — that is the point of the exercise. Then write the date beside them, because a census without a timestamp is a rumour, and the discipline of stamping it is worth more to you than the numbers.

Now derive four ratios and write what each one means in a sentence:

  edits ÷ articles                         1,370,633,170 / 7,240,827  =  189.29
  active share of registered                  262,745 / 54,539,294  =  0.4818 %
  active accounts per administrator               262,745 / 809  =  324.8
  articles per administrator                    7,240,827 / 809  =  8,950.3

The 0.4818 per cent is the one to sit with. Fewer than five in a thousand registered accounts did anything in the last month. That is not a scandal; it is the structural condition of every knowledge commons, and it is exactly what principle 2B is about.

Exercise 2.2 — Score it (3 hours)

Use the chapter's rubric. Write the score and the evidence for each of the twelve. The evidence is the assignment; the score is a byproduct.

  0  absent
  1  present in form only
  2  present and operating
  3  present, operating, and measurable from outside without permission

The chapter's score for English Wikipedia was 29 of 36 — 80.6 per cent, with a core (1B, 2A, 2B, 4B, 5, 6) of 16 of 18 — 88.9 per cent and a non-core of 13 of 18 — 72.2 per cent.

Disagree with it. Seriously — the instrument is more useful to you if you have argued with it once. Write down every place you would score differently and why. Two students scoring the same commons and landing on different totals is not a failure of the rubric; it is the reason inter-rater reliability has to be published, which is Exercise 3.2.

Exercise 2.3 — Convert points into bits (90 minutes)

This is the exercise that changes how you read every scorecard you meet afterwards.

     W  =  log2 [ P(principle present | endures) / P(present | fails) ]

     evidence  =  sum over i of  (score_i / 3) x W_i

For a score of 29, the sum of score/3 is 9.667. Compute the posterior probability of enduring at three assumed likelihood ratios, starting from prior odds of 0.4658 — a 31.78 per cent base rate:

  LR = 1.2        W = 0.2630      evidence  2.54 bits      P =  73.08 %
  LR = 2.0        W = 1.0000      evidence  9.67 bits      P =  99.74 %
  LR = 4.0        W = 2.0000      evidence 19.33 bits      P =  99.9997 %

Then write one paragraph answering: which of those three is right?

The honest answer is that nobody knows, because the likelihood ratio requires knowing how often each principle was present in the commons that died, and almost nobody has counted the dead. That single paragraph is the intellectual content of this whole chapter, and writing it yourself is how it sticks.

Exercise 2.4 — Find the base rate (2 hours)

Go and read the only kind of study that counts failures: a population census.

Schweik and English classified the whole SourceForge population:

  projects, 2009 census                    174,333       (107,747 in 2006)
  success in growth                         24,899       14.28 %
  abandoned in growth                       53,450       30.66 %
  determinate growth outcomes               78,349
  success share of determinate                            31.78 %

And TeBlunthuis, Shaw and Hill studied 740 wikis — the top one per cent by unique registered article editors, implying a population of about 74,000, of which 73,260 have never been examined.

Now compute the comparison that makes the whole problem visible:

  Cox's entire case base                        77 cases
  one digital-commons census               174,333 projects
  ratio                                     0.0442 %
  projects per Cox case                        2,264

Write one paragraph on what that ratio does and does not mean. It does not mean the commons literature is wrong — its cases are far more deeply coded than any census row. It means the two literatures are answering different questions, and only one of them has a denominator.

PART THREE — THE TERM PROJECT

Weeks nine to twelve: build the cohort nobody has built

This is the project, and it is the reason this chapter is worth a term.

The finding is available and has not been claimed. The likelihood ratios that would turn a commons score into a forecast require an inception cohort: cases coded at founding, blind to outcome, followed until each is alive or dead.

Exercise 3.1 — Build a birth register (two weeks)

Choose a population with a public creation date. Options that work:

Every repository on a public forge that crossed five distinct committers in a given month, two to five years ago.
Every wiki created on a given host in a given month, three or more years ago.
Every open standards working group chartered in a given year.
Every mutual aid group, housing co-operative or community energy scheme registered with a national regulator in a given year, where registers are public.

The register must be a birth register, not a hall of fame. If your selection criterion contains any measure of later success, you have rebuilt the fault this project exists to fix. Write your inclusion rule down before you look at a single case, and do not change it afterwards.

Target: 40 to 60 cases. The full study wants 230 — 73 survivors per group at 80 per cent power with a Bonferroni correction over eleven principles, at a 31.78 per cent base rate. You are not doing the full study. You are doing the pilot that shows it can be done, and a well-documented 50 is a genuine contribution.

Exercise 3.2 — Code them blind, twice (three weeks)

Two raters. Neither knows the outcome. Code each case on the twelve as they stood at founding, using only material dated before your t₀.

Then publish the disagreement. Report inter-rater reliability by principle, and name the principles you could not code reliably. Ratajczyk and colleagues wrote an entire paper on exactly this problem in commons coding; read it before you start rather than after.

This is the exercise where most of the learning happens, and the most valuable output is the list of principles that turned out to be uncodeable from the outside. That list is a finding.

Exercise 3.3 — Attach the outcomes (one week)

Now, and only now, look at what happened. Define alive and dead before you look — write the definition down and date it. Then build the 2×2 for each principle and compute the likelihood ratio you have been reasoning about all term.

                    present    absent
   endured             a          b
   died                c          d

           LR  =  [ a / (a+b) ]  /  [ c / (c+d) ]

Exercise 3.4 — Write it up honestly (one week)

Four sections and a table. State your denominator in the first paragraph — how many cases you started with, how many you could code, how many reached a determinate outcome, and how many you had to drop and why. A study that hides its attrition is an advertisement.

And report the principles where your ratio came out at or below one. Those are the most interesting results you will have, and they are the ones a nervous writer buries.

SELF-ASSESSMENT

Mark yourself honestly. The scale is the same as the rubric, which is deliberate.

	0	1	2	3
The original wording	Know the list	Can recite eight	Can state four in Ostrom's terms	Can name what each paraphrase dropped, and why it matters
The record	Heard of Cox et al.	Know it was supportive	Know the 91, the 77 and the 3.73	Can explain why the implicit studies scored higher and what that licenses
Eight into twelve	Aware of a reformulation	Can name the splits	Can name all three splits	Can explain the 4A/4B letter trap without notes
Configuration	Know the list is not additive	Can cite Baggio et al.	Know the core four	Can explain why 4B matters in forestry and not fisheries
Scoring	Understand the rubric	Scored one commons	Scored one and defended each score with evidence	Scored one and found where the rubric itself fails
Points into bits	Understand the distinction	Can run the calculation	Can run it and state the band	Can explain to a non-specialist what the score does not license, in one minute, without hedging
The base rate	Know it is unknown	Can name the two selection faults	Can distinguish selection-on-survival from selection-on-documentation	Have counted something nobody had counted
The cohort	Understand the design	Can size it	Built a birth register	Coded blind, reported reliability, published the dead

Three at the bottom row and you have done original work. Take it to whoever supervises you, because it is publishable and the field wants it.

THE ONE THING TO CARRY

The principles are excellent and the evidence behind them is real. They tell you where a commons is weak. They do not tell you whether it will live.

Somebody will hand you a scorecard — for a commons, a supplier, a country, a person — and will read a probability off it. When that happens, ask the one question this chapter exists to give you:

What is the likelihood ratio, and who measured it?

If the answer is a shrug, the score is a diagnosis. Use it as one. It is still worth having, and it is worth far more when nobody is pretending it is something else.

And one thing more, which is the reason this chapter sits where it does in the edition. The people who built the commons that failed are the ones who would make the calibration possible, and nobody has ever gone and found them. Every study in the field so far has been written about institutions that were still standing when the researcher arrived. That is not a criticism of anybody; it is simply the shape of what has been easy to reach. What has changed is that a great many commons now leave a complete public record of their entire life, including the end of it, and that record does not require a field trip or a translator or a grant. It requires somebody to decide that the dead are worth counting.

You are as well placed to make that decision as anyone who has ever held a chair in this field, and better placed than most of them were, because the data arrived after they had already chosen their questions. Choose yours knowing it is there.