When Does a Value Turn Against Its Purpose?

On the difference between legitimately sacrificing a value and its turning against those it was meant to protect

Values do not live alone. Justice competes with peace, freedom with security, truthfulness with mercy. Anyone who demands a system in which no value ever gives way to another is demanding a system that decides nothing. Every serious decision involves a sacrifice of some kind: one genuine value steps back so that another can step forward.

Yet history knows situations that resemble this on the surface and differ from it in substance: a value raised in the name of protecting people, which ends by harming the very people it was raised to protect, or by serving others. The difficulty is that the two look alike at the moment of decision. In both, the decision-maker says that he is placing this value above that one because the common good requires it. How, then, do we distinguish a legitimate competition between two genuine values from a value that has begun to work against its purpose?

I have addressed this phenomenon more broadly in an earlier lecture that has not yet been published. Here I ask a narrower question, one closer to examination: faced with a decision that ranks one value above another, what signs were available at the moment of decision, not after the harm occurred, that could alert us that we may be at the beginning of a reversal rather than a legitimate competition?

I do not claim that these signs amount to a complete definition. A reversal is not established merely because the choice caused harm; every choice harms something. What I propose is that the likelihood of reversal becomes serious when three signs coincide: the person harmed is denied an objection that can have effect; the constraints of necessity, proportionality and review are absent; and the distribution of cost is skewed in a way not offset by an independent mechanism that recognizes those harmed and reckons with their claims. I set these signs against three cases to which they do not apply equally, and the comparison is preliminary, nothing more.

A first case: amnesty with a condition

When South Africa emerged from apartheid, it faced an explicit dilemma between two genuine values: justice required prosecuting those who had committed crimes, and civil peace required that prosecutions not become a new war. The Promotion of National Unity and Reconciliation Act of 1995 established the Truth and Reconciliation Commission and empowered it to grant amnesty for acts committed between 1960 and 1994.

The amnesty was not unconditional. The Act required that the act be associated with a political objective and that the applicant make full disclosure of all relevant facts, and it obliged the Amnesty Committee, in applications that went to a hearing, to notify victims of the hearing and of their right to attend and testify. The Amnesty Committee received 7,115 applications. Most were decided in chambers without a public hearing because they did not concern gross violations, and most were refused.

Relatives of victims, together with the organization AZAPO, then challenged the constitutionality of the amnesty before the Constitutional Court, arguing that it confiscated their right to go to court, since the Act extinguished not only criminal but also civil liability. The Court heard the challenge and upheld the Act in July 1996. That judgment does not prove that the victims' rights were adequately protected. What it proves is narrower: that the challenge could be brought, that it was decided by a reasoned judgment, and that the Act itself imposed a condition on the perpetrator, namely that he stand and recount what he had done.

I do not say that the settlement succeeded. The Commission recommended annual grants to victims for six years; after a delay, the government settled for a single payment of thirty thousand rand to some twenty-one thousand registered victims. Many victims still regard the process as a settlement made at their expense. But that very criticism is available because the settlement was exposed: written, conditional, challenged, and with a known party bearing its cost.

A second case: protection by coercion

On 2 May 1927 the United States Supreme Court decided Buck v. Bell by eight votes to one, upholding a 1924 Virginia statute that authorized the compulsory sterilization of those the state deemed "feeble-minded." Public health is a legitimate value in itself, but the judgment did not rest on it in the abstract. It rested on a particular interpretation of it, eugenics, with the class judgments it carried about who deserved to have children. Justice Oliver Wendell Holmes relied on the precedent of compulsory vaccination: if the state may compel people to be vaccinated to protect the community, it may cut a woman's Fallopian tubes for the same end. He closed with his notorious sentence: "Three generations of imbeciles are enough."

Laws of this kind existed, at one time or another, in thirty-three states, and more than sixty thousand Americans were sterilized under them, from the first law, in Indiana in 1907, to the mid-1970s. As for Carrie Buck herself, the historian Paul Lombardo concluded that the evidence of her feeble-mindedness was almost nonexistent, and that her daughter, the "third generation" in Holmes's phrase, earned acceptable grades at school.

Carrie Buck's case did pass through the courts, and she had a lawyer. But Lombardo showed that her lawyer, Irving Whitehead, was one of the closest friends of Aubrey Strode, counsel for the other side, and that after losing he went to the board of the Colony to report that the case was in excellent shape to proceed to the Supreme Court. The eugenics archive of the Cold Spring Harbor Laboratory and the journal of the Virginia State Bar both record that he presented no evidence and called no witnesses on her behalf at the first trial. Friendship alone proves nothing, but counsel's relationship with the opposing party, together with his documented conduct at trial, raises serious doubts about the independence and effectiveness of the defense. The procedure satisfied its form, while its substance remained seriously in doubt.

A third case: amnesty without a disclosure condition

So that the difference between the first two cases does not appear to rest solely on the kind of right involved, an amnesty in one case and a body in the other, I add a case of the same kind as the first. On 18 April 1978 the military junta governing Chile issued Decree Law 2191, granting amnesty for crimes committed under the state of siege between 11 September 1973 and 10 March 1978. It is a written amnesty, like the South African one, and it too was justified in the name of national reconciliation. It covered perpetrators, accomplices and those who concealed the crimes; its wording also covered opponents convicted by military courts; and it excluded a range of ordinary crimes, such as rape, robbery with violence or breaking, and drug trafficking, as well as one particular case then pending before the military courts. But it required neither disclosure nor confession, it established no mechanism to examine each case individually, and its most prominent effect was to shield from prosecution those who had committed violations under the authority that issued it.

Here it helps to distinguish between moments. When the decree was issued, there was no parliament to debate it: legislative power lay with the junta itself. After its issue, the courts applied it and closed cases, and the Chilean Supreme Court upheld its constitutionality in August 1990, after the return of civilian rule. I do not judge here the extent of the Chilean judiciary's independence in practice during those years. As for the 2006 judgment of the Inter-American Court of Human Rights in Almonacid-Arellano, which held that applying the decree to prevent accountability violates the American Convention on Human Rights, it is a later assessment, not evidence that objection was possible in 1978.

This shows that being written down does not by itself make the difference. Two written amnesties, both in the name of peace, diverge clearly on the first two signs. One was adopted by an elected parliament, required disclosure, opened its hearings to victims, and could be challenged before an existing constitutional court. The other was issued by the authority under which the crimes were committed, with no disclosure condition and no body before which its constitutionality could be tested other than a judiciary operating under that same authority.

Three signs

It is not enough to say that the first settlement is a competition of values and the other two cases are reversals because their outcomes were worse. We need signs tied to the moment of decision, and a rule for each sign, announced before the cases are examined. Before the signs there must be a rule for time itself, and it requires a distinction. What existed at the moment of decision is one thing; what was available to its parties is another. The first is the features of the decision and of the bodies applying it as they objectively were, in its text, in the composition of those bodies and in the records of their proceedings, even if historians uncovered them only decades later. The second is what the person harmed, and anyone wishing to object on their behalf, actually knew or could have known. A document kept in the archive of one party existed in the first sense, but it was not an available warning in the second. If we want the signs to serve as a tool that alerts the decision-maker and his opponents at the time, only the second counts. If we want to test them against history, we code both and keep them apart. What happened after the decision, challenges decided, judgments issued, compensation paid, does not enter into the judgment of a sign; it is used to verify it: was the path that seemed open actually open?

The first sign: did objection retain any possible effect? This is not measured by the outcome, by whether the objector won, or the judgment becomes circular. It is measured by indicators prior to the outcome: representation independent of the other party, access to the evidence, a real path of challenge, a reasoned response to the objection, and the possibility of later review. These indicators do not all fall at one moment or on one level, so we must distinguish between objecting to the decision itself, before its adoption or by challenging it afterward, and objecting to its application in each case. The capacity to obtain redress is the question of the third sign, not the first. In South Africa, the Act was adopted by an elected parliament after negotiations that, according to the Commission's final report, included wide public debate; but that does not prove that victims had any influence on the design of the settlement itself. At the moment of its adoption, a challenge to its constitutionality existed and was known, because the interim Constitution of 1993 had established a Constitutional Court to review the constitutionality of acts of parliament. Objection in individual applications was provided for in the text of the Act itself, through victims' right to attend and testify at hearings, even though most applications were decided without a hearing because they did not concern gross violations. What came afterward is later verification: the challenge to the Act was brought and decided by a reasoned judgment in 1996, even though the applicants lost; and eight applications for judicial review of Amnesty Committee decisions had been brought before the courts by the time the report was prepared, two of them successfully, and were referred back to the Committee. In Carrie Buck's case, the first of the indicators, independence of representation, was already compromised in the trial itself in 1924. But this belongs to what existed, not to what was available: it came to light only decades later, and Carrie Buck could not have known it. What was available at the time was the text of the statute, which left the definition of "feeble-mindedness" to the body applying it, and that is a matter for the second sign. In Chile, before the decree there was no parliament before which to object, and after it there was no body independent of the junta before which its constitutionality could be tested; both facts were known to everyone. The Supreme Court's confirmation of its constitutionality in 1990 is later verification showing that the obstacle was not merely theoretical.

The second sign: is the retreat constrained? Constraint here means three things: that the means be connected to the end in a way that can be demonstrated; that the harm be necessary and proportionate, with no less harmful alternative; and that the decision be subject to review. The South African amnesty was limited by a time period, by the nature of the act, and by a condition, full disclosure, and it could be subjected to judicial review. But the existence of these constraints does not establish necessity and proportionality. That the amnesty was a term of the negotiated settlement, written into the postamble of the interim Constitution of 1993, shows that the negotiators regarded it as a condition of the transition; it does not by itself demonstrate that no less harmful alternative existed. And proportionality is contested: the Act extinguished civil liability along with criminal liability, and put in their place only reparations that could be recommended. The success of this case on the second sign is therefore partial and disputed, and I claim nothing more. The Virginia statute remained in force until its repeal in 1974, limited only by a loose definition of "feeble-mindedness" controlled by the body applying it. The Chilean decree was limited by a time frame and a list of excluded crimes, but it tied the amnesty to no price paid by the beneficiary and subjected its application to no case-by-case examination.

The third sign: on whom does the cost fall, who reaps the benefit, and were those harmed recognized and their claims reckoned with? This is a sign about the distribution of harm, not a definition of reversal. Because almost every public decision benefits one group and costs another, a skewed distribution is not enough to satisfy it. I set a rule for it before examining the cases: the sign is satisfied when two conditions coincide. The first is that the group promised protection bears a permanent cost not linked to the end by any means open to examination, or that the holder of decision-making power and those under his command are the first direct beneficiaries. The second is that neither the decision itself, nor anything in place when it was adopted, provides a mechanism independent of the decision-maker that recognizes those harmed, assesses their claims and repairs them. The existence of such a mechanism on paper is not enough. One must ask about its independence: who appoints its members, whether they answer to the decision-maker or to those who benefit from the decision, and whether a political body reviews its decisions. And one must ask about its effectiveness: whether it can bind, or only recommend. What is created years after the decision does not count in its favor at the moment of adoption, even if it later mitigates the harm. The two alternatives in the first condition are two forms of a single imbalance: the cost is placed on those promised protection, or the benefit goes to those who hold the decision. The cost of a decision may fall on the very group promised protection without that being a reversal. A proportionate quarantine restricts the patient himself in the name of health, but it is temporary, has a rationale that can be examined, and is subject to review by a body other than the one that imposed it.

By this rule the sign is satisfied in Virginia. Sterilization was presented as serving both the interest of the person sterilized and the interest of society; its permanent cost then fell on those subjected to it. Sterilization began at the request of the superintendent of the institution in which they lived and was decided by its board, with a subsequent right of appeal to the courts, and the defense before the courts is seriously in doubt. It is satisfied in Chile: the most prominent beneficiaries were those who had committed violations under the authority that issued the amnesty, and the decree established no mechanism that recognized the victims or examined their claims. In South Africa the first condition is partially satisfied, since the beneficiaries included members of the parties that negotiated the settlement on both sides. The result on the second condition is mixed, and the judgment on it rests on the text of the Act at the time of its adoption. The Act itself established an Amnesty Committee chaired by a judge, provided that it act independently of any party or government, barred the parent Commission from reviewing its decisions, and opened its hearings to victims. Recognition of victims, in the sense of hearing their voices and confronting the perpetrator before them, was therefore available at the moment of decision, although this was independence provided for in law, and I do not judge its extent in practice. But the assessment and repair of their claims did not lie with a body that could bind: a decision of the Amnesty Committee took effect and released the perpetrator, while the Committee on Reparation and Rehabilitation could, under the text of the Act, only make recommendations to the President, who would consider them in order to make recommendations to Parliament. Giving victims a voice is not the same as substantive redress, and the third sign here is neither decisively absent nor fully satisfied. What came later confirms this coding rather than creating it. In the very Constitutional Court judgment that upheld the Act, Justice Didcott described the trade-off as replacing the tangible reality of legal claims, where these existed, with the indefinite prospect of reparations; and the compensation actually paid fell short of what the Commission recommended. The difference that remains between this case and Chile is that here the cost was written down, announced and open to challenge, in the name of another, explicitly stated value: peace.

The signs, then, do not apply equally to the three cases. They appear to coincide in Virginia and in Chile, each in its own way. In South Africa the first was present at the moment of decision, the second partially, and the third was mixed: procedural recognition of victims without binding repair of their claims. I do not conclude from this that the signs explain what happened better than other accounts, for the three cases were chosen after their outcomes were known; they are examples that illustrate the idea, not evidence that proves it. Nor is any one of the signs sufficient by itself. But when they coincide, they become a warning that obliges the decision-maker to answer the objections with reasons, to record them, and to open a path for reviewing the decision. Suspending the decision until the objections are answered is the right response only when its harm is irreversible and the cost of delay does not match that gravity. In transitional decisions, delay itself may be very costly; sterilization cannot be undone.

The South African case may reveal a third analytical category between legitimate competition and reversal: a declared settlement with procedural conditions, which acknowledges its cost without guaranteeing its repair. It might more precisely be called an exposed settlement with an unrepaired cost. The condition imposed on the perpetrator, disclosure, is a procedural condition, not a price that repairs the victim; the unrepaired cost remained with the victims. This is a descriptive category for an institutional structure, not a normative verdict that acquits or condemns. A case enters it only if three things are present: the cost is declared in the text of the decision; those harmed receive procedural recognition in which their voices are heard; and their claims have no binding repair. Without these conditions, the category becomes a refuge for every mixed case. A case's entering it does not by itself establish the settlement's legitimacy or rule out its reversal. And if it is indeed a category in its own right, the question it raises is not only whether the value turned against its purpose, but who carries the debt after the decision has passed.

How can this hypothesis be tested?

What I have presented here is a preliminary comparison of three selected cases, not a test. The hypothesis admits two tests, which should not be confused.

The first is predictive: do the signs precede the violation and warn of it? To keep this test from becoming circular, the outcome must be defined without reference to the signs. I propose that a decision be counted as having turned against its purpose if later records establish three things together. The first is harm reaching a level of gravity announced before coding, being one of two things: irreversible injury to bodily integrity, or denial of redress for grave crimes. By denial of redress I mean that both criminal prosecution and civil action for the act are closed to the person harmed, without binding reparation in their place, throughout a time horizon announced before coding, ten years for instance, beginning from the date the decision takes effect for the beneficiary, that is, from the grant of an individual amnesty or the entry into force of a general one. Where partial reparation exists without litigation, it is coded in a separate category, partial denial, and is counted neither as denial nor as its absence. The mere restriction of a right, or an adverse effect that accompanies every harsh decision, does not count as harm. The second is that this harm falls on the group promised protection or on a defined group of those harmed, not on individuals incidentally. The third is that the harm be attributed to the decision to an announced degree: direct attribution when the harm is what the decision provides for in its text, such as the sterilization it authorizes or the exemption from prosecution it grants, in which case proof of application to the group suffices; and indirect attribution when the harm passes through other intermediaries, in which case it can be established only by comparison with cases in which the decision was not taken, and is given lower weight. Direct legal effect should not be confused with direct historical effect. The legal effect in all three cases here is direct: the text authorizes sterilization, and the text extinguishes prosecution. But whether the measured harm occurred because of the text, and not because of decisions by prosecutors or courts or obstacles that existed before it, requires independent tracing for each case, and the need is greater in Chile, because other obstacles to prosecution, such as the jurisdiction of military courts, may have had an independent effect that the text of the decree does not settle. I exclude two things from this definition. The first is a finding that the decision was unlawful, because that is the normative question that comes below; were it included, the two tests would become one. The second is the extent to which the declared aim was achieved, because that is a third question, measured separately: a decision may achieve its aim and harm those promised protection at the same time. A serious test would then require an independent sample of decisions that rank values against each other, not chosen because of their outcomes; coding of the signs under the rule for time stated above, separating what was available to the parties to the decision from what existed but became known only later, carried out by coders who do not know how the decisions ended, so that no warning is read into a decision that became visible only later; and coding of outcomes by a separate team with no access to the coding of the signs. The hypothesis then fails if the coincidence of available signs is not more frequent before violations than before other outcomes, and it is weakened if exposed, constrained settlements, in which the indicators of effective objection were present, repeatedly turned against their purpose all the same. I want to know of such cases, if they exist.

The second is normative: was the decision permissible at all? The signs do not settle this. A decision may be taken through a rigorous procedure, with challenge, reasons and review, and still be unjust in its substance. And a decision may succeed in achieving its declared aim without that establishing its legitimacy. The signs warn of danger; they do not exempt us from moral judgment.

Two objections

The first objection: these are procedural signs, and procedures can be satisfied in form. Carrie Buck's case is itself an example: a lawyer, hearings, and a judgment from the highest court. The objection is well taken, and that is why I did not make the first sign "Was a hearing held?" but "Was representation independent, was the challenge real, was the response reasoned?" The Whitehead episode shows the difference: the form was satisfied, and the substance is seriously in doubt. That is why the three signs work together, not each one alone.

The second objection: this is judgment with hindsight. In the 1920s eugenics enjoyed influential support among scientific and legal elites. That is true as a description of the climate, but it does not cancel the signs. The consensus was not complete: Justice Pierce Butler dissented from the majority in the same case, and there were opponents of these policies outside the Court. And the data for the signs were available in 1927 to anyone willing to see them: the text of the statute left the definition of "feeble-mindedness" to the body applying it, and those subject to it were inmates of state institutions. That Carrie Buck's objection found no one to carry it forcefully is a historical conclusion made likely by what Lombardo documented decades later, not a judgment reached at the time, and so I do not rest my reply to this objection on it. What was lacking was not that the data be available, but that they be regarded as mattering.

Conclusion

Not every sacrifice of a value is a betrayal of it. A society that will not let one value give way to another cannot decide anything. But there is a difference between a value that steps back while remaining present, its voice heard and its price counted, and a value expelled from the reckoning, after which its sister value is invoked to commit what it was established to prevent. Between them lies a zone in which it is not enough for the voice to be heard if the debt is not paid.

Anyone facing a decision that ranks one value above another can, before asking whether the ranking is right, ask three questions. If the person harmed objects, who carries the objection, and who is obliged to answer it? What constrains this decision and makes it open to review? And who will pay its price, who will reap its fruit, and who other than the decision-maker has the power to settle the claim of the one who pays?

Sources

1. Buck v. Bell, 274 U.S. 200 (1927); Embryo Project Encyclopedia, Arizona State University, "Buck v. Bell (1927)". https://supreme.justia.com/cases/federal/us/274/200/ https://embryo.asu.edu/pages/buck-v-bell-1927

2. Paul A. Lombardo, Three Generations, No Imbeciles: Eugenics, the Supreme Court, and Buck v. Bell (Baltimore: Johns Hopkins University Press, 2008).

3. Paul Lombardo, "Carrie Buck's lawyer working against her interests", DNA Learning Center, Cold Spring Harbor Laboratory. https://dnalc.cshl.edu/view/15236-Carrie-Buck-s-lawyer-working-against-her-interests-Paul-Lombardo.html

4. DNA Learning Center, "Irving Whitehead, the lawyer appointed to represent Carrie Buck, called no witnesses at her trial in Amherst County, Virginia"; DNA Learning Center, "Irving Whitehead, still image with audio"; Virginia Lawyer (Virginia State Bar), June/July 2013, p. 30. https://dnalc.cshl.edu/view/12179-Irving-Whitehead-the-lawyer-appointed-to-represent-Carrie-Buck-called-no-witnesses-at-her-trial-in-Amherst-County-Virigina.html https://dnalc.cshl.edu/view/15831-Irving-Whitehead-still-image-with-audio.html https://virginialawyer.vsb.org/va-lawyer-jun-jul-2013/page-30

5. Eugenics Archive, "Eugenic Sterilization Laws". https://eugenicsarchive.org/html/eugenics/essay8text.html

6. Claude Moore Health Sciences Library, University of Virginia, "Eugenics: Three Generations, No Imbeciles", Epilogue. https://exhibits.hsl.virginia.edu/eugenics/5-epilogue

7. Promotion of National Unity and Reconciliation Act 34 of 1995 (South Africa), assented to 19 July 1995, sections 1(2), 4(f), 5(e), 17, 19(3), 19(4), 20(7), 25(1)(b), 27 and 36; Constitution of the Republic of South Africa Act 200 of 1993, Chapter 7 (Constitutional Court) and Postamble "National Unity and Reconciliation". https://www.saflii.org/za/legis/num_act/ponuara1995477/

8. Truth and Reconciliation Commission of South Africa Report, Volume 1, Chapter 5, paras. 9 and 59; Volume 6, Section 1, Chapter 2, paras. 26 and 69, and Chapter 3, para. 4. https://www.polity.org.za/polity/govdocs/commissions/1998/trc/1chap5.htm https://www.gov.za/sites/default/files/gcis_document/201409/trc0.pdf

9. Truth and Reconciliation Commission, "Amnesty applications" (statistics as at 15 November 1999), statement of 9 December 1999. https://www.polity.org.za/polity/govdocs/pr/1999/pr1209d.html

10. Azanian Peoples Organization (AZAPO) and Others v President of the Republic of South Africa and Others, [1996] ZACC 16, 1996 (4) SA 671 (CC), 25 July 1996, paras. 3, 46–47, and para. 65 (Didcott J). https://www.saflii.org/za/cases/ZACC/1996/16.html

11. Decreto Ley N° 2.191, Concede amnistía, Diario Oficial, 19 April 1978. https://www.cooperativa.cl/noticias/site/artic/20140911/asocfile/20140911191144/dl_2191_19_abr_1978.pdf

12. Inter-American Commission on Human Rights, Report No. 34/96, Cases 11.228 et al. (Chile), 15 October 1996, paras. 4 and 105. https://hrlibrary.law.umn.edu/cases/1996/chile34-96.htm

13. Equipo Nizkor, on the August 1990 judgment of the Supreme Court of Chile upholding the constitutionality of Decree Law 2191. https://derechos.org/nizkor/chile/doc/amnistia4.html

14. Inter-American Court of Human Rights, Almonacid-Arellano et al. v. Chile, Judgment of 26 September 2006, Series C No. 154, paras. 119, 122, 145 and 151; Official summary of the judgment; Monitoring Compliance with Judgment, Order of 18 November 2010, considering paragraph 13 (operative paragraphs 5 and 6). https://www.corteidh.or.cr/docs/casos/articulos/seriec_154_esp.pdf https://www.corteidh.or.cr/docs/resumen/almonacid_arellano.pdf https://www.corteidh.or.cr/docs/supervisiones/almonacid_18_11_10.pdf

15. Amar Murray, "The Pathology of the Balance" (Arabic: «باثولوجيا الميزان»), lecture in the School of Values series, 2026 (unpublished).

16. Amar Murray, "When Good Principles Are Not Enough". https://amarmurray.com/en/writing/when-principles-are-not-enough/