La milla verde (I): el caso contra John Coffey

Portada de la resena de «La milla verde» (The Green Mile, 1999).

Idioma / Language: Español · English

Primera entrega de una saga de tres artículos sobre la película de Frank Darabont. Un hombre encontrado junto al río con dos niñas muertas en brazos, un juicio del que nadie se acuerda y una silla eléctrica al final del pasillo: la anatomía de un error judicial, contada con lo que hoy sabemos sobre visión de túnel, discapacidad intelectual, defensa de oficio y ciencia forense sin validar.

Read in English ▶

Autora: Manuela González

Por qué se necesitan tres artículos para analizar esta película

La milla verde (The Green Mile, Frank Darabont, 1999) dura ciento ochenta y nueve minutos y cabe en una sola frase: en la Luisiana de 1935, los funcionarios del corredor de la muerte de la penitenciaría de Cold Mountain reciben a un condenado que no ha hecho lo que dice la sentencia. Adapta la novela que Stephen King publicó por entregas en 1996, en seis fascículos mensuales, y obtuvo cuatro candidaturas a los Óscar, entre ellas la de mejor película y la de actor secundario para Michael Clarke Duncan. Tom Hanks interpreta a Paul Edgecomb, el jefe de vigilantes del bloque E, y el pasillo de linóleo verde que da nombre a todo es el trayecto que un preso recorre una sola vez.

Desde la criminología y las ciencias forenses, la película ofrece bastante más de lo que suele reconocérsele. Hay en ella un error judicial completo, con todos sus ingredientes clásicos; hay una institución que mata por procedimiento y unos funcionarios que la sirven; y hay dos ejecuciones filmadas con un detalle técnico poco frecuente en el cine. Tratar todo eso en un solo texto obligaría a despachar cada asunto en dos párrafos, así que hemos preferido dividirlo en tres entregas. Esta primera se ocupa del caso: cómo se llega a condenar a muerte a un hombre inocente. La segunda entrará en el bloque E, en quienes trabajan allí y en lo que la espera hace con quien aguarda. La tercera se ocupará de las formas de matar que el Estado ha ido ensayando, de la electrocución al gas y a la inyección, y del papel que la raza del acusado juega en ese desenlace.

Conviene el aviso habitual: para hablar del caso hay que contar la película entera, incluido su final.

Lo que vio la partida de búsqueda

El relato del crimen llega al espectador de segunda mano, como llegan casi siempre las cosas en esta película. Dos niñas, las hermanas Detterick, desaparecen de noche del porche donde dormían. El padre encuentra sangre, la alarma corre por el condado y se organiza una batida con perros y escopetas. Al amanecer, la partida llega a la orilla de un río y encuentra a un hombre enorme, sentado, con los dos cuerpos en los brazos, la ropa empapada de sangre y llorando. Cuando le preguntan, dice que no pudo evitarlo, que lo intentó.

Ahí se acabó la investigación. No en sentido figurado: en la película no hay una sola escena que sugiera que alguien buscó después una explicación distinta de la que tenía delante. Esa imagen, que es material probatorio de primer orden, se trató como si fuera una conclusión ya redactada.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
El ingreso en el bloque E. En la película, el proceso ya ha terminado; el espectador solo conoce el caso por lo que cuentan quienes vigilan la puerta. · Fotograma de La milla verde (The Green Mile, 1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

La distinción entre indicio y prueba es el primer contenido de cualquier manual de investigación criminal, y consiste justamente en no dar ese salto. Un hombre sosteniendo dos cadáveres es un hecho que admite varias explicaciones: que los matara, que los encontrara, que llegara a tiempo de intentar algo y demasiado tarde para lograrlo, o que llegara después que otro. El trabajo de investigación consiste en formular esas hipótesis y buscar activamente los datos que puedan descartar cada una. Ninguna de esas cosas ocurre en el caso de John Coffey. La escena se cerró con la primera hipótesis disponible y todo lo que vino después —el atestado, la acusación, el juicio— se limitó a decorarla.

Visión de túnel, o el momento en que un caso deja de investigarse

Ese mecanismo tiene nombre técnico. Keith Findley y Michael Scott lo describieron con detalle en un trabajo de 2006 que sigue siendo la referencia sobre el asunto: la visión de túnel es la tendencia de los actores del sistema penal a filtrar selectivamente la información para reforzar una conclusión adoptada de antemano. Empieza en el investigador que decide pronto quién es el culpable, sigue en el fiscal que solo pide las diligencias que confirman su tesis, y llega al juez y al jurado, que reciben un expediente ya podado. Su ingrediente principal es el sesgo de confirmación, esa comodidad mental que hace que los datos favorables a nuestra hipótesis parezcan sólidos y los desfavorables, anecdóticos.

Lo relevante para lo que aquí interesa es que la visión de túnel no requiere mala fe. Ninguno de los hombres que detiene a Coffey es un villano; son vecinos aterrorizados que acaban de ver a dos niñas muertas. Esa es exactamente la razón por la que el fenómeno resulta tan difícil de combatir, y por la que los protocolos actuales insisten en medidas estructurales —revisión por equipos distintos, registro de las hipótesis descartadas y de sus motivos, obligación de investigar también lo que exculpa— en lugar de confiar en la buena voluntad de nadie.

Las cifras del National Registry of Exonerations, el registro que documenta las exoneraciones en Estados Unidos desde 1989, dan la medida del problema. A comienzos de 2026 el registro contabilizaba más de 3.700 personas exoneradas y por encima de treinta y cinco mil años de prisión cumplidos por quienes no habían cometido el delito. Solo en 2024 hubo 147 exoneraciones, y en 104 de ellas —el 71 %— se documentó mala praxis de funcionarios públicos, policiales o judiciales. Entre los factores que concurren en esos casos, el registro señala el falso testimonio o la falsa acusación en el 72 %, la defensa deficiente en el 33 %, la prueba forense errónea o presentada de forma engañosa en el 29 % y la falsa confesión en el 15 %. Casi ningún caso tiene una sola causa; lo habitual es que se acumulen.

En España tenemos un espejo bastante exacto. Dolores Vázquez fue condenada en 2001 por un tribunal del jurado como autora del asesinato de Rocío Wanninkhof, sin una sola prueba biológica que la vinculara con el crimen, sobre un conjunto de indicios interpretados a través de la antipatía que despertaba su carácter y su vida privada, y con una presión mediática que hoy resulta difícil de leer sin incomodidad. Cumplió 519 días de prisión. La verdad llegó por accidente en 2003: tras el asesinato de Sonia Carabantes en Coín, el perfil genético hallado en aquella escena coincidió con restos recuperados en el caso Wanninkhof y condujo a Tony Alexander King, un británico con antecedentes por agresión sexual en su país. Aquel error judicial no lo corrigió una revisión del razonamiento del tribunal; lo corrigió un segundo crimen.

Un acusado que no puede defenderse

John Coffey mide dos metros, pesa más de cien kilos y tiene una discapacidad intelectual evidente. No sabe leer más allá de su nombre, no recuerda dónde nació, responde a las preguntas con el eco de la última que le han hecho y llama «jefe» a todo el que lleva uniforme. En el relato, esa condición no aparece jamás como un problema jurídico. Nadie evalúa si comprende la acusación, si puede ayudar a su defensa o si entiende qué significa la pena que le espera.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Las cadenas de un condenado a muerte de 1935. La talla y la fuerza del acusado funcionaron en el juicio como un argumento más. · Fotograma de La milla verde (The Green Mile, 1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

Hoy, ese sería el primer asunto que un tribunal tendría que resolver, y son en realidad dos preguntas distintas que conviene no mezclar. Una es la capacidad procesal: si el acusado comprende el procedimiento y puede participar en él con un mínimo de sentido. La otra es la imputabilidad: si en el momento de los hechos podía comprender la ilicitud de lo que hacía y actuar conforme a esa comprensión. Un acusado puede ser plenamente imputable y a la vez incapaz de seguir un juicio, y al revés. En el caso de Coffey no consta que nadie se hiciera ninguna de las dos preguntas.

El derecho estadounidense tardó sesenta y siete años en cerrar la puerta que la película deja abierta. En Atkins contra Virginia (2002), el Tribunal Supremo declaró inconstitucional ejecutar a personas con discapacidad intelectual, por entender que había cambiado el criterio social sobre lo que constituye un castigo cruel. La sentencia, sin embargo, dejó en manos de cada estado la definición de quién entra en esa categoría, y el resultado fue una colección de criterios dispares que el propio Tribunal tuvo que corregir después: en Hall contra Florida (2014) rechazó los umbrales rígidos de cociente intelectual que ignoraban el error de medida de las pruebas, y en Moore contra Texas (2017, con una segunda resolución en 2019) descartó los criterios legos que un estado había construido al margen del conocimiento clínico, insistiendo en que la valoración debe atender a los déficits de conducta adaptativa y no a las habilidades conservadas del acusado.

En el ordenamiento español la discusión se plantea en otros términos, porque no hay pena capital que discutir desde 1978. La anomalía o alteración psíquica que impide comprender la ilicitud del hecho exime de responsabilidad conforme al artículo 20.1 del Código Penal, con la posibilidad de imponer medidas de seguridad orientadas al tratamiento. Y en el plano procesal, la Convención de las Naciones Unidas sobre los derechos de las personas con discapacidad, ratificada por España, obliga en sus artículos 12 y 13 a proporcionar apoyos para el ejercicio de la capacidad jurídica y ajustes de procedimiento —lenguaje accesible, facilitadores, tiempos adaptados— para garantizar el acceso efectivo a la justicia. Un acusado como Coffey tendría hoy derecho a que alguien se asegurara de que entiende lo que ocurre en la sala.

«No pude evitarlo, jefe»

La frase que Coffey repite al ser detenido es literalmente cierta: no pudo evitar que las niñas murieran, llegó tarde. Todo el que la escucha entiende otra cosa. En el expediente pasó a ser una confesión, y funcionó como el broche del caso.

La investigación sobre confesiones falsas describe con precisión ese tipo de trampa. Saul Kassin y Lawrence Wrightsman propusieron hace décadas una clasificación que sigue vigente y distingue tres formas: la confesión voluntaria, que nadie ha provocado; la confesión por acatamiento, en la que el sospechoso cede a la presión del interrogatorio para terminar con la situación aun sabiendo que es inocente; y la confesión interiorizada, la más inquietante, en la que la persona acaba creyendo que cometió el delito. Gisli Gudjonsson desarrolló además los instrumentos que hoy se emplean para medir la sugestibilidad interrogativa —la tendencia a incorporar información sugerida por el entrevistador y a modificar las respuestas ante la presión—, y sus escalas mostraron algo que la práctica confirma a diario: quienes tienen discapacidad intelectual, menores de edad o trastornos mentales puntúan sistemáticamente más alto.

El peso real del fenómeno se ve en los números. Alrededor del 29 % de las personas exoneradas por pruebas de ADN documentadas por el Innocence Project habían confesado, se habían autoinculpado o habían hecho declaraciones incriminatorias sobre un delito que no cometieron. La creencia de que nadie confiesa un crimen ajeno es una de las intuiciones más extendidas y peor fundadas del sistema penal.

Las salvaguardas que hoy se consideran mínimas nacieron precisamente de casos así: grabación audiovisual íntegra del interrogatorio y no solo de la declaración final, presencia efectiva de abogado, prohibición de presentar al sospechoso pruebas falsas, y evaluación previa de la comprensión cuando hay indicios de discapacidad. Ninguna de ellas existía en la Luisiana de 1935, y varias siguen sin ser obligatorias en muchos lugares.

Un abogado que ya tenía la respuesta

Hay una escena breve y demoledora en la segunda mitad de la película. Paul Edgecomb visita al abogado que defendió a Coffey en el juicio, un letrado de oficio que le recibe en el jardín de su casa y le explica, con la tranquilidad de quien repite algo obvio, que su cliente era como un perro de buena raza que un día muerde sin motivo y al que solo cabe sacrificar. Ese hombre representó a un acusado en un procedimiento donde estaba en juego su vida sin haber creído en ningún momento en su inocencia.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Coffey y Edgecomb en la celda. La película dedica horas a la conversación que la investigación del caso nunca tuvo. · Fotograma de La milla verde (The Green Mile, 1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

El detalle histórico es que, en 1935, aquel acusado ya tenía derecho a abogado, y por una razón que enlaza directamente con la tercera entrega de esta saga. En Powell contra Alabama (1932), el caso de los llamados «chicos de Scottsboro» —nueve adolescentes negros acusados de violar a dos mujeres blancas en un tren de Alabama, juzgados en cuestión de días y condenados a muerte casi todos ellos—, el Tribunal Supremo estableció que en los procesos capitales el Estado debe designar abogado al acusado que no puede pagarlo y darle tiempo real para preparar la defensa. Tres años después, en la Luisiana de la película, la garantía existía sobre el papel. Coffey tuvo abogado. Lo que no tuvo fue defensa.

Medio siglo más tarde, en Strickland contra Washington (1984), el Tribunal Supremo fijó el criterio con el que se juzga si una defensa fue tan mala como para anular una condena: hay que demostrar dos cosas a la vez, que la actuación del letrado quedó por debajo de un estándar objetivo de razonabilidad y que, de haber actuado correctamente, existe una probabilidad razonable de que el resultado hubiera sido otro. La doctrina se ha criticado desde el primer día por su enorme deferencia hacia el abogado y por la dificultad práctica de acreditar el segundo requisito, y no es casual que la defensa inadecuada siga apareciendo en un tercio de las exoneraciones documentadas.

En el marco europeo el enfoque es distinto y algo más exigente en su formulación. El artículo 6.3 del Convenio Europeo de Derechos Humanos reconoce el derecho a defenderse mediante abogado, y el Tribunal de Estrasburgo lleva desde Artico contra Italia (1980) repitiendo que el Convenio protege derechos prácticos y efectivos, de modo que la simple designación de un letrado no basta si este no desempeña una defensa real. En España, ese contenido se encauza por el derecho a la tutela judicial efectiva y a la defensa del artículo 24 de la Constitución.

Lo que hoy diría la criminalística

Supongamos por un momento que el caso Detterick se instruye hoy. Lo primero que cambiaría es lo que ocurrió antes de que llegara nadie con formación: una partida de decenas de vecinos, con perros y armas, recorriendo durante horas el mismo terreno en el que estaban las víctimas. El principio de intercambio que formuló Edmond Locard funciona en ambas direcciones, y una escena pisoteada por sesenta personas pierde los vestigios del autor y se llena, encima, de otros nuevos. La primera medida de cualquier protocolo actual es acordonar y restringir el acceso, y se toma justamente porque el instinto colectivo empuja a hacer lo contrario.

Con la escena preservada, el caso se resolvería probablemente en el laboratorio. El delito incluía agresión sexual, de modo que habría material biológico del autor: un perfil genético obtenido del semen, de los restos subungueales de las víctimas o del ADN de contacto en la ropa. Ese perfil se compararía con el del detenido y con las bases de datos policiales. La entomología forense permitiría acotar el intervalo postmortem mínimo a partir de la colonización por insectos, con los márgenes y las limitaciones que la propia disciplina reconoce. Y la distribución de las manchas de sangre en el peto de Coffey admitiría un análisis orientado a distinguir la transferencia pasiva —el resultado de abrazar dos cuerpos ya sangrantes— de la proyección producida por los golpes.

Ese último punto pide una advertencia honesta, porque el cine ha hecho mucho daño en este terreno. La interpretación de patrones de manchas de sangre no es una lectura infalible. El estudio más riguroso realizado hasta la fecha, dirigido por Austin Hicklin y publicado en 2021, sometió a 75 analistas en ejercicio a 192 patrones representativos del trabajo real y obtuvo un 11,2 % de respuestas erróneas en las muestras de origen conocido, además de un 7,8 % de conclusiones que contradecían las de otros analistas sobre el mismo patrón. Es una herramienta útil cuyos límites conviene declarar en sala.

Que todo esto no es un ejercicio teórico lo demuestra un caso real muy parecido al de la película. En 1985, Kirk Bloodsworth fue condenado a muerte en Maryland por la violación y el asesinato de una niña de nueve años, sobre la base de identificaciones de testigos y sin una sola prueba física que lo vinculara al crimen. En 1993, el análisis del semen recuperado en la escena demostró que no era suyo y salió en libertad: fue el primer condenado a muerte de Estados Unidos exonerado por ADN. El perfil genético identificó años después al autor real, que cumplía condena por otro delito en esa misma prisión.

Perros, mordeduras y pelos: la ciencia que no lo era

El otro movimiento es el inverso, y resulta menos cómodo de explicar: no todo lo que se presenta ante un tribunal con bata blanca merece llamarse ciencia. El informe que el National Research Council estadounidense publicó en 2009 sobre el estado de las ciencias forenses fue el primer aviso serio, y en 2016 el consejo asesor científico de la Presidencia volvió sobre el asunto con un criterio más severo: de todas las disciplinas de comparación examinadas, únicamente el ADN de aportante único, las mezclas sencillas de ADN y, con matices, la identificación de huellas dactilares latentes acreditaban validez científica demostrada mediante estudios empíricos.

El caso más citado es el de las marcas de mordedura. Aquel informe de 2016 concluyó que no existía ni un solo estudio que estableciera la validez del método y que la literatura disponible apuntaba en dirección contraria: los examinadores ni siquiera coinciden de forma fiable en si una lesión es una mordedura humana. Ray Krone fue condenado en Arizona en 1992 por el asesinato de una camarera, con las marcas dentales como prueba prácticamente única; pasó diez años en prisión, dos de ellos en el corredor de la muerte, y en abril de 2002 el ADN identificó al verdadero autor y lo dejó en libertad. Fue la persona número cien exonerada del corredor de la muerte estadounidense desde 1976.

El pelo humano dio un disgusto todavía mayor, porque el problema estaba dentro del laboratorio de referencia. En 2015, el FBI y el Departamento de Justicia hicieron público el resultado parcial de una revisión de los casos en que sus examinadores habían declarado sobre comparación microscópica de cabellos: de los 268 juicios revisados en los que ese testimonio se usó para incriminar al acusado, en 257 —el 96 %— los peritos habían hecho afirmaciones que excedían lo que la técnica permite sostener. En al menos 35 de esos casos el acusado fue condenado a muerte, y en 33 se detectaron errores. Conviene precisar qué significa aquí «error»: los peritos expresaron sus conclusiones con una rotundidad que la técnica no permitía sostener, al margen de que el origen atribuido a cada pelo fuese o no el correcto.

Y quedan los perros, que en la película llevan a la partida hasta la orilla del río. Un perro puede rastrear un olor sobre el terreno y encontrar a una persona; eso lo hace bien y es una herramienta de búsqueda útil. Cosa distinta es la llamada rueda de reconocimiento olfativa, en la que se pretende que el animal señale al culpable entre varias muestras de olor. En Texas, un ayudante del sheriff llamado Keith Pikett realizó con sus perros más de dos mil identificaciones de esa clase a lo largo de veinte años; el Innocence Project de aquel estado documentó en 2009 la práctica en un informe cuyo título hablaba directamente de pseudociencia, varias condenas basadas en esas ruedas fueron revocadas y un experto británico en unidades caninas que revisó una grabación del procedimiento lo describió como el método policial más primitivo que había visto nunca. La diferencia entre buscar a alguien y señalar a un culpable es exactamente la diferencia entre una herramienta y una prueba.

Wild Bill: el autor estaba en la celda de al lado

El giro del argumento coloca al verdadero responsable a pocos metros del condenado. William Wharton, «Wild Bill», ingresa en el bloque E por otros delitos y se comporta desde el primer día como el problema disciplinario que es: agresivo, imprevisible, sin la menor consideración por las consecuencias, disfrutando del efecto que produce en los demás. Solo al final se sabe que fue él quien entró en el porche de los Detterick.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Vigilante y condenado, la pareja sobre la que descansa el relato. · Fotografía promocional de La milla verde (The Green Mile, 1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

La tentación de ponerle una etiqueta clínica al personaje hay que resistirla en parte. La escala de psicopatía de Robert Hare, la PCL-R, se aplica mediante entrevista estructurada y revisión exhaustiva del expediente, y no se puntúa a alguien a partir de lo que muestra una película. Dicho esto, el guion reúne en Wharton un catálogo bastante reconocible de rasgos: manipulación instrumental, ausencia total de remordimiento, control conductual deficiente, búsqueda de estimulación y una crueldad que no persigue ningún beneficio.

El detalle criminalísticamente más fino del argumento es el modo en que se cometió el crimen. Wharton amenaza a cada niña con matar a su hermana si grita, y con eso consigue que ninguna de las dos haga ruido mientras se las lleva. Eso describe una conducta planificada, con conocimiento de la víctima y control de la escena, que es precisamente lo que la literatura sobre homicidio sexual asocia a los perfiles con puntuaciones altas en el componente interpersonal y afectivo de la psicopatía; los estudios sobre homicidio de menores han encontrado además que quienes matan a niños puntúan de forma significativamente más alta en esa escala que quienes cometen abusos sin llegar al homicidio, y la combinación de sadismo sexual y psicopatía se asocia a conductas de control en la escena y a la reiteración.

Hay una consecuencia de los errores judiciales de la que se habla poco, y esta película la escenifica sin subrayarla: mientras un inocente cumple la condena, el autor real continúa su vida. En el conjunto de casos del Innocence Project en los que el ADN permitió identificar después al verdadero responsable, esos autores fueron condenados por alrededor de ciento cincuenta delitos violentos posteriores, entre ellos decenas de agresiones sexuales y más de treinta homicidios, cometidos durante los años en que otra persona pagaba por lo que ellos habían hecho. No hace falta ir a Estados Unidos para verlo: Sonia Carabantes tenía diecisiete años cuando fue asesinada en 2003, cuatro años después de la muerte de Rocío Wanninkhof y dos después de la condena de una mujer inocente.

Por qué sigue siendo un buen caso práctico

La película acierta en lo más difícil, que es mostrar que para condenar a un inocente no hace falta ningún conspirador. Hay un funcionario sádico en el bloque E, Percy Wetmore, y de él nos ocuparemos en la segunda entrega, pero el error judicial de Coffey no lo comete él. Lo cometen un padre destrozado, unos vecinos asustados, unos agentes convencidos, un abogado desganado y un jurado que vio lo que cualquiera habría visto. Todos ellos estaban razonablemente seguros de tener razón. Esa es la lección más incómoda y la que conviene llevarse.

Donde el argumento carga las tintas es en el terreno donde no podía hacer otra cosa. El espectador sabe que Coffey es inocente porque la película le concede un don sobrenatural que lo demuestra; en un caso real esa certeza no existe nunca, y el sistema tiene que funcionar con la duda. De hecho, el argumento contiene su propia refutación, porque el don que prueba la inocencia de Coffey ante nosotros no le sirvió de nada ante el tribunal. Lo único que habría podido salvarlo era un procedimiento que buscara activamente su inocencia, y de eso no hubo ni rastro.

En la próxima entrega cruzaremos la puerta del bloque E para mirar la institución por dentro: el coste psíquico que el trabajo tiene para quienes ejecutan, el deterioro que produce la espera en quien aguarda su turno, la violencia del funcionario que sabe que nadie va a tocarle y el problema, poco tratado, de los condenados que renuncian a recurrir y piden que se cumpla la sentencia.

The Green Mile (I): the case against John Coffey

Leer en español ▶

The first of three articles on Frank Darabont’s film. A man found by the river with two dead girls in his arms, a trial nobody remembers and an electric chair at the end of the corridor: the anatomy of a wrongful conviction, told through what we now know about tunnel vision, intellectual disability, court-appointed defence and unvalidated forensic science.

By Manuela González

Three articles for a very long film

The Green Mile (Frank Darabont, 1999) runs for one hundred and eighty-nine minutes and fits into a single sentence: in 1935 Louisiana, the officers of the death house at Cold Mountain Penitentiary take delivery of a prisoner who did not do what the sentence says he did. It adapts the novel Stephen King published in serial form in 1996, in six monthly instalments, and it earned four Academy Award nominations, including Best Picture and Best Supporting Actor for Michael Clarke Duncan. Tom Hanks plays Paul Edgecomb, the head guard of E Block, and the green linoleum corridor that gives the film its title is the walk a prisoner takes only once.

Seen from criminology and the forensic sciences, the film offers a good deal more than it is usually given credit for. It contains a complete wrongful conviction, with every one of its classic ingredients; an institution that kills by procedure and the men who serve it; and two executions filmed with a technical detail rarely seen in cinema. Covering all of that in one text would mean disposing of each subject in two paragraphs, so we have split it into three parts. This first one deals with the case: how an innocent man ends up sentenced to death. The second will step inside E Block, into the people who work there and what the waiting does to the man who waits. The third will look at the methods of killing the State has tried out over time, from electrocution to gas and to the needle, and at the part the defendant’s race plays in that outcome.

The usual warning applies: discussing the case means giving away the whole film, ending included.

What the search party saw

The account of the crime reaches the audience second-hand, as almost everything does in this film. Two little girls, the Detterick sisters, disappear at night from the porch where they were sleeping. Their father finds blood, the alarm spreads through the county and a party sets out with dogs and shotguns. At dawn they reach a riverbank and find an enormous man sitting there, holding both bodies in his arms, his clothes soaked in blood, weeping. Asked what happened, he says he could not help it, that he tried.

That is where the investigation stopped. Not figuratively: there is not a single scene suggesting that anyone later looked for an explanation other than the one in front of them. That image, which is evidence of the first order, was treated as a conclusion already written out.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Arrival on E Block. By the time the film starts, the trial is over; the audience learns about the case only through the men who guard the door. · Still from The Green Mile (1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

The distinction between an indication and proof is the first thing any criminal investigation manual teaches, and it consists precisely in not taking that leap. A man holding two bodies is a fact open to several explanations: that he killed them, that he found them, that he arrived in time to try something and too late to succeed, or that he arrived after someone else. Investigative work consists of framing those hypotheses and actively looking for the data that might rule each one out. None of that happens in John Coffey’s case. The scene was closed around the first available hypothesis, and everything that followed — the police report, the charge, the trial — merely decorated it.

Tunnel vision, or the moment a case stops being investigated

The mechanism has a technical name. Keith Findley and Michael Scott described it in detail in a 2006 paper that remains the reference work on the subject: tunnel vision is the tendency of actors in the criminal justice system to filter information selectively so as to reinforce a conclusion reached in advance. It starts with the investigator who decides early who the culprit is, continues with the prosecutor who only orders the enquiries that confirm the theory, and reaches the judge and the jury, who receive an already pruned file. Its main ingredient is confirmation bias, the mental comfort that makes evidence favourable to our hypothesis look solid and unfavourable evidence look anecdotal.

What matters here is that tunnel vision requires no bad faith. None of the men who arrest Coffey is a villain; they are terrified neighbours who have just seen two dead children. That is exactly why the phenomenon is so hard to fight, and why current protocols insist on structural safeguards — review by separate teams, a written record of discarded hypotheses and the reasons for discarding them, a duty to investigate exculpatory lines as well — rather than trusting anyone’s good will.

The figures from the National Registry of Exonerations, which has documented exonerations in the United States since 1989, give the measure of the problem. By early 2026 the Registry recorded more than 3,700 people exonerated and over thirty-five thousand years served in prison by people who had not committed the crime. In 2024 alone there were 147 exonerations, and in 104 of them — 71 % — official misconduct by police, prosecutors or judges was documented. Among the contributing factors, the Registry records perjury or false accusation in 72 % of cases, inadequate legal defence in 33 %, false or misleading forensic evidence in 29 % and false confession in 15 %. Hardly any case has a single cause; they tend to pile up.

Spain has a fairly exact mirror image. In 2001, Dolores Vázquez was convicted by a jury court of the murder of Rocío Wanninkhof without a single piece of biological evidence linking her to the crime, on a set of circumstantial indications read through the dislike aroused by her personality and her private life, and under media pressure that is uncomfortable to revisit today. She served 519 days. The truth arrived by accident in 2003: after the murder of Sonia Carabantes in Coín, the genetic profile found at that scene matched material recovered in the Wanninkhof case and led to Tony Alexander King, a British man with previous convictions for sexual assault in his own country. That miscarriage of justice was not corrected by any review of the court’s reasoning; it was corrected by a second murder.

A defendant who cannot defend himself

John Coffey is two metres tall, weighs well over a hundred kilos and has an obvious intellectual disability. He cannot read beyond his own name, does not remember where he was born, answers questions by echoing the last one put to him and calls everyone in uniform «boss». In the story, that condition never appears as a legal issue. Nobody assesses whether he understands the charge, whether he can assist his defence or whether he grasps what the sentence awaiting him means.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
The chains of a 1935 death-row prisoner. The defendant’s size and strength worked at trial as one more argument against him. · Still from The Green Mile (1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

Today that would be the first matter a court had to settle, and it is in fact two separate questions that should not be blurred. One is fitness to stand trial: whether the defendant understands the proceedings and can take part in them with some minimum of sense. The other is criminal responsibility: whether, at the time of the acts, he could understand the wrongfulness of what he was doing and act on that understanding. A defendant may be fully responsible and yet unable to follow a trial, and the other way round. In Coffey’s case there is no sign that anyone asked either question.

American law took sixty-seven years to close the door this film leaves open. In Atkins v. Virginia (2002) the Supreme Court held it unconstitutional to execute people with intellectual disability, on the view that society’s standards about what counts as cruel punishment had shifted. The ruling, however, left it to each state to define who falls into that category, and the result was a collection of divergent criteria that the Court itself had to correct later: in Hall v. Florida (2014) it rejected rigid IQ cut-offs that ignored the standard error of measurement, and in Moore v. Texas (2017, with a second ruling in 2019) it struck down the lay criteria one state had built outside clinical knowledge, insisting that the assessment must attend to deficits in adaptive behaviour rather than to the defendant’s preserved strengths.

In Spanish law the discussion takes a different shape, since there has been no death penalty to argue about since 1978. A mental anomaly or disorder that prevents a person from understanding the wrongfulness of the act exempts them from responsibility under article 20.1 of the Criminal Code, with the possibility of imposing security measures aimed at treatment. Procedurally, the United Nations Convention on the Rights of Persons with Disabilities, ratified by Spain, requires in articles 12 and 13 that support be provided for the exercise of legal capacity and that procedural accommodations be made — accessible language, facilitators, adapted timeframes — to guarantee effective access to justice. A defendant like Coffey would today have the right to have someone make sure he understands what is happening in the courtroom.

«I couldn’t help it, boss»

The sentence Coffey repeats when he is arrested is literally true: he could not help the girls dying, he arrived too late. Everyone who hears it understands something else. In the file it became a confession, and it worked as the clasp that closed the case.

Research on false confessions describes that kind of trap precisely. Saul Kassin and Lawrence Wrightsman proposed decades ago a classification that still holds and distinguishes three forms: the voluntary confession, which nobody prompted; the compliant confession, in which the suspect yields to interrogation pressure to bring the situation to an end while knowing he is innocent; and the internalised confession, the most disturbing, in which the person comes to believe he committed the crime. Gisli Gudjonsson went on to develop the instruments used today to measure interrogative suggestibility — the tendency to take on information suggested by the interviewer and to change answers under pressure — and his scales showed something daily practice confirms: people with intellectual disability, minors and those with mental disorders score consistently higher.

The weight of the phenomenon shows in the numbers. Around 29 % of the people exonerated by DNA testing in the cases documented by the Innocence Project had confessed, incriminated themselves or made damaging statements about a crime they did not commit. The belief that nobody confesses to someone else’s crime is one of the most widespread and worst-founded intuitions in criminal justice.

The safeguards now regarded as minimal grew out of cases like these: full audiovisual recording of the interrogation and not only of the final statement, the effective presence of a lawyer, a ban on presenting suspects with fabricated evidence, and a prior assessment of comprehension where there are signs of disability. None of them existed in 1935 Louisiana, and several are still not mandatory in many places.

A lawyer who already had the answer

There is a brief, devastating scene in the second half of the film. Paul Edgecomb visits the lawyer who defended Coffey at trial, a court-appointed attorney who receives him in his garden and explains, with the calm of a man repeating the obvious, that his client was like a well-bred dog that one day bites for no reason and can only be put down. That man represented a defendant in a proceeding where his life was at stake without ever having believed in his innocence.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Coffey and Edgecomb in the cell. The film spends hours on the conversation the investigation never had. · Still from The Green Mile (1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

The historical detail is that in 1935 the defendant already had a right to counsel, for a reason that connects directly with the third part of this series. In Powell v. Alabama (1932), the case of the so-called Scottsboro Boys — nine Black teenagers accused of raping two white women on an Alabama train, tried within days and almost all sentenced to death — the Supreme Court held that in capital proceedings the State must appoint counsel for a defendant who cannot afford one and give that lawyer real time to prepare. Three years later, in the Louisiana of the film, the guarantee existed on paper. Coffey had a lawyer. What he did not have was a defence.

Half a century later, in Strickland v. Washington (1984), the Supreme Court set the standard for deciding whether a defence was bad enough to void a conviction: two things must be shown at once, that counsel’s performance fell below an objective standard of reasonableness and that, had counsel performed adequately, there is a reasonable probability the outcome would have been different. The doctrine has been criticised from day one for its enormous deference to counsel and for the practical difficulty of establishing the second limb, and it is no accident that inadequate defence still appears in a third of documented exonerations.

The European framework takes a different and somewhat more demanding approach in its wording. Article 6.3 of the European Convention on Human Rights recognises the right to defend oneself through legal assistance, and since Artico v. Italy (1980) the Strasbourg Court has repeated that the Convention protects rights that are practical and effective, so that merely appointing a lawyer is not enough if that lawyer provides no real defence. In Spain this content runs through the right to effective judicial protection and defence in article 24 of the Constitution.

What forensic science would say today

Suppose for a moment that the Detterick case were investigated today. The first thing to change would be what happened before anyone trained arrived: a party of dozens of neighbours, with dogs and weapons, walking for hours over the very ground where the victims lay. Edmond Locard’s exchange principle works in both directions, and a scene trampled by sixty people loses the offender’s traces while filling up with fresh ones. The first measure in any current protocol is to cordon off and restrict access, precisely because collective instinct pushes the other way.

With the scene preserved, the case would probably be settled in the laboratory. The crime included sexual assault, so there would be biological material from the offender: a genetic profile from semen, from the victims’ fingernail scrapings or from contact DNA on clothing. That profile would be compared with the detainee’s and with police databases. Forensic entomology would allow a minimum post-mortem interval to be estimated from insect colonisation, with the margins and limitations the discipline itself acknowledges. And the distribution of the bloodstains on Coffey’s overalls would allow an analysis aimed at distinguishing passive transfer — the result of holding two already bleeding bodies — from spatter produced by blows.

That last point calls for an honest caveat, because cinema has done real damage here. Bloodstain pattern interpretation is not an infallible reading. The most rigorous study to date, led by Austin Hicklin and published in 2021, put 75 practising analysts through 192 patterns representative of real casework and found 11.2 % erroneous responses on samples of known origin, plus 7.8 % of conclusions that contradicted those of other analysts on the same pattern. It is a useful tool whose limits ought to be stated in court.

That none of this is a theoretical exercise is shown by a real case remarkably close to the film. In 1985 Kirk Bloodsworth was sentenced to death in Maryland for the rape and murder of a nine-year-old girl, on the basis of witness identifications and without a single piece of physical evidence linking him to the crime. In 1993, analysis of semen recovered at the scene showed it was not his and he walked free: he was the first death-row prisoner in the United States exonerated by DNA. The genetic profile identified the real offender years later, a man serving a sentence for another crime in that same prison.

Dogs, bite marks and hairs: the science that was not

The opposite move is less comfortable to explain: not everything presented to a court in a white coat deserves to be called science. The report published in 2009 by the American National Research Council on the state of the forensic sciences was the first serious warning, and in 2016 the President’s Council of Advisors on Science and Technology returned to the subject with a stricter yardstick: of all the comparison disciplines examined, only single-source DNA, simple DNA mixtures and, with qualifications, latent fingerprint identification could show scientific validity established by empirical studies.

The most cited example is bite mark comparison. That 2016 report concluded there was not a single study establishing the validity of the method and that the available literature pointed the other way: examiners cannot even agree reliably on whether an injury is a human bite. Ray Krone was convicted in Arizona in 1992 of the murder of a bartender, with dental marks as practically the only evidence; he spent ten years in prison, two of them on death row, and in April 2002 DNA identified the real killer and set him free. He was the hundredth person exonerated from American death row since 1976.

Human hair produced an even worse shock, because the problem sat inside the reference laboratory. In 2015 the FBI and the Department of Justice released the partial results of a review of cases in which their examiners had testified about microscopic hair comparison: of the 268 trials reviewed in which that testimony was used to incriminate the defendant, in 257 — 96 % — the experts had made statements going beyond what the technique can support. In at least 35 of those cases the defendant was sentenced to death, and errors were found in 33. It is worth being precise about what «error» means here: the examiners voiced their conclusions with a certainty the technique could not support, regardless of whether the origin attributed to a given hair was correct.

Then there are the dogs, which in the film lead the party to the riverbank. A dog can follow a scent over ground and find a person; it does that well and it is a useful search tool. A quite different matter is the so-called scent lineup, in which the animal is supposed to point out the culprit among several odour samples. In Texas, a sheriff’s deputy named Keith Pikett carried out more than two thousand such identifications with his dogs over twenty years; the Innocence Project of Texas documented the practice in a 2009 report whose title spoke plainly of junk science, several convictions based on those lineups were overturned, and a British police dog-handling expert who reviewed a recording of the procedure described it as the most primitive evidential police procedure he had ever witnessed. The difference between finding someone and identifying a culprit is exactly the difference between a tool and a piece of evidence.

Wild Bill: the offender was in the next cell

The turn of the plot places the real culprit a few metres from the condemned man. William «Wild Bill» Wharton arrives on E Block for other crimes and behaves from day one like the disciplinary problem he is: aggressive, unpredictable, wholly indifferent to consequences, enjoying the effect he produces on others. Only at the end does it emerge that he was the man who climbed onto the Dettericks’ porch.

Imagen de la película «La milla verde» (The Green Mile, 1999), de Frank Darabont.
Guard and condemned man, the pair the whole story rests on. · Promotional photograph from The Green Mile (1999), dir. Frank Darabont. © Castle Rock Entertainment / Warner Bros.

The temptation to pin a clinical label on the character must partly be resisted. Robert Hare’s psychopathy scale, the PCL-R, is applied through a structured interview and an exhaustive file review, and nobody is scored on the basis of what a film shows. That said, the script gathers in Wharton a fairly recognisable catalogue of traits: instrumental manipulation, complete absence of remorse, poor behavioural control, sensation seeking and a cruelty that pursues no benefit at all.

The finest forensic detail in the plot is the way the crime was committed. Wharton threatens each girl with killing her sister if she screams, and in that way keeps both of them silent while he carries them off. That describes planned conduct, with knowledge of the victims and control of the scene, which is precisely what the literature on sexual homicide associates with high scores on the interpersonal and affective component of psychopathy; studies of child homicide have also found that those who kill children score significantly higher on that scale than those who abuse without killing, and the combination of sexual sadism and psychopathy is associated with controlling behaviour at the scene and with repetition.

There is a consequence of wrongful convictions that is rarely discussed, and this film stages it without underlining it: while an innocent person serves the sentence, the real offender carries on with his life. Across the Innocence Project cases in which DNA later identified the true perpetrator, those offenders were convicted of around one hundred and fifty further violent crimes, including dozens of sexual assaults and more than thirty homicides, committed during the years someone else was paying for what they had done. There is no need to look to the United States: Sonia Carabantes was seventeen when she was murdered in 2003, four years after the death of Rocío Wanninkhof and two years after an innocent woman was convicted for it.

Why it remains a good teaching case

The film gets the hardest thing right, which is to show that convicting an innocent man requires no conspirator. There is a sadistic officer on E Block, Percy Wetmore, and we shall deal with him in the second part, but Coffey’s wrongful conviction is not his doing. It is the work of a shattered father, some frightened neighbours, some convinced officers, a listless lawyer and a jury that saw what anyone would have seen. All of them were reasonably sure they were right. That is the most uncomfortable lesson, and the one worth taking away.

Where the story overplays its hand is in the one place it could not do otherwise. The audience knows Coffey is innocent because the film grants him a supernatural gift that proves it; in a real case that certainty never exists, and the system has to work with doubt. In fact the plot contains its own refutation, because the gift that proves Coffey’s innocence to us was of no use to him before the court. The only thing that could have saved him was a procedure that actively looked for his innocence, and there was no trace of that.

In the next instalment we shall step through the door of E Block to look at the institution from the inside: the psychological cost the work carries for those who carry out executions, the deterioration produced by waiting in the man whose turn is coming, the violence of the officer who knows nobody will touch him, and the little-discussed problem of condemned prisoners who waive their appeals and ask for the sentence to be carried out.

Bibliografía / References

Amendt, J., Richards, C. S., Campobasso, C. P., Zehner, R., & Hall, M. J. R. (2011). Forensic entomology: Applications and limitations. Forensic Science, Medicine, and Pathology, 7(4), 379-392. https://doi.org/10.1007/s12024-010-9209-2

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5.ª ed., texto revisado). https://doi.org/10.1176/appi.books.9780890425787

Atkins v. Virginia, 536 U.S. 304 (2002). https://supreme.justia.com/cases/federal/us/536/304/

Artico c. Italia, n.º 6694/74 (Tribunal Europeo de Derechos Humanos, 13 de mayo de 1980). https://hudoc.echr.coe.int/eng?i=001-57424

Federal Bureau of Investigation. (2015, 20 de abril). FBI testimony on microscopic hair analysis contained errors in at least 90 percent of cases in ongoing review. https://www.fbi.gov/news/press-releases/fbi-testimony-on-microscopic-hair-analysis-contained-errors-in-at-least-90-percent-of-cases-in-ongoing-review

Findley, K. A., & Scott, M. S. (2006). The multiple dimensions of tunnel vision in criminal cases. Wisconsin Law Review, 2006(2), 291-397. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=911240

Gudjonsson, G. H. (2018). The psychology of false confessions: Forty years of science and practice. Wiley. https://doi.org/10.1002/9781119315636

Hall v. Florida, 572 U.S. 701 (2014). https://supreme.justia.com/cases/federal/us/572/701/

Hare, R. D. (2003). Manual for the Hare Psychopathy Checklist-Revised (2.ª ed.). Multi-Health Systems.

Hicklin, R. A., Winer, K. R., Kish, P. E., Parks, C. L., Chapman, W., Dunagan, K., Richetelli, N., Epstein, E. G., Ausdemore, M. A., & Busey, T. A. (2021). Accuracy and reproducibility of conclusions by forensic bloodstain pattern analysts. Forensic Science International, 325, 110856. https://doi.org/10.1016/j.forsciint.2021.110856

Innocence Project. (s. f.). DNA exonerations in the United States (1989-2020). Recuperado el 23 de agosto de 2026 de https://innocenceproject.org/dna-exonerations-in-the-united-states/

Innocence Project of Texas. (2009). Dog scent lineups: A junk science injustice. https://innocenceproject.org/news/new-report-examines-unreliable-dog-scent-lineups/

Kassin, S. M. (2017). False confessions: How can psychology so basic be so counterintuitive? American Psychologist, 72(9), 951-964. https://doi.org/10.1037/amp0000195

Kassin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., & Redlich, A. D. (2010). Police-induced confessions: Risk factors and recommendations. Law and Human Behavior, 34(1), 3-38. https://doi.org/10.1007/s10979-009-9188-6

Moore v. Texas, 581 U.S. 1 (2017). https://supreme.justia.com/cases/federal/us/581/15-797/

National Registry of Exonerations. (2025). Annual report 2024. University of California Irvine, University of Michigan Law School & Michigan State University College of Law. https://exonerationregistry.org/publications

National Research Council. (2009). Strengthening forensic science in the United States: A path forward. The National Academies Press. https://doi.org/10.17226/12589

Naciones Unidas. (2006). Convención sobre los derechos de las personas con discapacidad. https://www.un.org/esa/socdev/enable/documents/tccconvs.pdf

Powell v. Alabama, 287 U.S. 45 (1932). https://supreme.justia.com/cases/federal/us/287/45/

President’s Council of Advisors on Science and Technology. (2016). Forensic science in criminal courts: Ensuring scientific validity of feature-comparison methods. Executive Office of the President. https://obamawhitehouse.archives.gov/sites/default/files/microsites/ostp/PCAST/pcast_forensic_science_report_final.pdf

Stefanska, E. B., Longpré, N., & Beauregard, E. (2019). Sexual sadism and psychopathy in sexual homicide offenders: An exploration of their associates in a clinical sample. International Journal of Offender Therapy and Comparative Criminology, 63(13), 2337-2358. https://doi.org/10.1177/0306624X19837048

Strickland v. Washington, 466 U.S. 668 (1984). https://supreme.justia.com/cases/federal/us/466/668/

Créditos de las imágenes / Image credits

Todas las imágenes de este artículo son fotogramas o material promocional de La milla verde (The Green Mile, Frank Darabont, 1999), © Castle Rock Entertainment / Warner Bros. Se reproducen en número limitado y a baja resolución al amparo del derecho de cita y del uso legítimo con fines de crítica, reseña y comentario. Todos los derechos sobre la obra pertenecen a sus titulares. Fuente de las imágenes: The Movie Database (image.tmdb.org).

All images in this article are stills or promotional material from The Green Mile (Frank Darabont, 1999), © Castle Rock Entertainment / Warner Bros. They are reproduced in limited number and at low resolution under the right of quotation and fair use for the purposes of criticism, review and comment. All rights in the work belong to their owners. Image source: The Movie Database (image.tmdb.org).

Comentarios

Deja una respuesta

Tu dirección de correo electrónico no será publicada. Los campos obligatorios están marcados con *