Mostly here because I was active a long time ago, but this is interesting enough to make an account again. If I use language that doesn't fit the lingo, that's why.
So you know if you want to bother reading-My conclusion is that I'm pretty sure you're wrong in ways that are fun and useful to discuss!
First off, it's absolutely relevant that the accused's knife isn't unique. If the knife it unique it selects for them specifically out of the suspect pool of everyone-it's not perfect evidence, it's possible that they lost it, but in most worlds where it's a unique knife they're the one who uses it on someone.
Following that, and this is contextual to the story and setting, the prior probability of someone owning a knife in his community (New York young man from a vaguely Hispanic community) is high. If the specific knife is thus common enough that it's possible to find a copy within a block or so, the chance that any suspect would have access to that knife becomes pretty reasonable. I'd vaguely estimate we're supposed to estimate a greater than 90% probably that any kid of his ethnicity and gender owns a knife, and at that point we're looking at hundreds of suspects who we know could have committed the crime with a perfect copy of the murder weapon because we know one was within walking distance. We know at least two of those knives existed near the murder, so let's spitball that the chance of anyone else using that knife is about equal to the chance of the suspect using the knife. 50-50. It's much more likely that the suspect had his knife than anyone else, but there are so many other people who could have committed the crime that I don't feel comfortable saying he's much more likely than everyone.
(In reality the ownership rate of knives was likely lower among Hispanic teens in 1950's America, the media exaggerated it, but people might borrow murder weapons, so we're into speculation land).
Second, the flaws in the eyewitness testimony are sufficient to basically completely obviate any of the witnesses as interrogators of truth. Eyewitnesses in perfect health and alacrity make major mistakes. There's too much evidence to summarize, but we're talking on the order of 75% chance of correct identification at absolute best. Arguably eyewitness testimony has basically no impact on probability estimates before we start adding in reasons to doubt it. In this context the stated circumstances-that the witness is elderly and saw the murder in the dark through a train-are actually sufficient to make it highly unreliable before we talk about glasses, which are actually just a red herring. Eyewitness testimony is so bad that they should have rejected it before they learned the witness even might be blind! If I was forced to give a prior probably of how likely that eyewitness was likely to pick the right suspect out of a lineup of all suspects in those circumstances, it'd be on the order of one in a thousand. There are likely nearly that many young people of vaguely correct build in a few block radius, and I have nearly zero confidence they could pick one out from another in the stated circumstance. I'm being generous by several orders of magnitude if they have vision problems.
Eyewitness testimony is that awful. It's the source of most miscarriages of justice, and it's arguably simply a mistake to use it in murder trials at all-the correct move is probably to ban it because it's more prejudicial to juries than evidential, even if a glimmer of scientific validity can be gleamed from it by an expert.
Hence the two/three pieces of reasonable evidence are-Knife, and fight with victim (whom was his dad). The fight is not great evidence, but specific threats to family members precede a lot of murders, believe it or not.
The prior probably that kid is guilty because the victim was his dad is around 10% (yes, really), the fight and lack of other familial suspects might increase that to 20%, and as I said I don't consider the knife as raising the odds more than a coin toss. The eyewitness testimony is a rounding error.
I'd hence conservatively put the kids chance of being guilty at ~70%. And that's completely unacceptable for a murder trial. That's barely a preponderance of evidence.
Legally, the correct move at this point is for the jury to walk out and admit everything they've discussed openly, at which point the entire case will be thrown out for a mistrial. Technically Juror 12 could have forced that himself by just saying that he saw an identical knife on the street, which is only a criminal (well...highly legally dubious) act when he smuggles it into the room. That's not good for the boy, because the idea that only one man in twelve was willing to murder him through legal maleficence for being Hispanic in 1950's new York is laughable, but it's the legal option. Juror 12, knowing that, is probably morally justified in pushing for an acquittal anyway.
As for shaming the racist juror into silence-shame is a powerful motivator. Racists, particularly modern racists, are often fundamentally social and physical cowards too, and they just bowled over him (mostly illegally) rather than relying on him to actually admit fault and chance his vote, which is more realistic to the psychology involved. Also fully justified, of course.
We can likely reject your alternative hypothesis because of the prior probably that someone who wanted to rig a trial could do so. There are tens of thousands of people whom might be called for jury duty (accounting for racial and gender discrimination...) on a given day in 1950's New York. Even knowing the day, hundreds are then called, of which 14 get put on a 12 man jury with two substitutes. You can go find the real numbers to do the math if you want, but back of the envelope, without a way to rig that system you're looking at a chance of less than 1/10000 if you knew that the juror wanted to rig the trial ahead of time. Rigging the system to influence juror selection is rare enough to have an infinitesimal prior probability if you know someone wants to. At least as a rogue juror. It's even rare in movie logic.
It's still possible to compromise a jury, but only after it's selected, and that is also extremely rare; if you're sure juror 12 is compromised he's remarkable composed for someone either approached for a massive bribe or whose family is being held hostage, which is how such things were actually done the few times it shows up.
Mostly here because I was active a long time ago, but this is interesting enough to make an account again. If I use language that doesn't fit the lingo, that's why.
So you know if you want to bother reading-My conclusion is that I'm pretty sure you're wrong in ways that are fun and useful to discuss!
First off, it's absolutely relevant that the accused's knife isn't unique. If the knife it unique it selects for them specifically out of the suspect pool of everyone-it's not perfect evidence, it's possible that they lost it, but in most worlds where it's a unique knife they're the one who uses it on someone.
Following that, and this is contextual to the story and setting, the prior probability of someone owning a knife in his community (New York young man from a vaguely Hispanic community) is high. If the specific knife is thus common enough that it's possible to find a copy within a block or so, the chance that any suspect would have access to that knife becomes pretty reasonable. I'd vaguely estimate we're supposed to estimate a greater than 90% probably that any kid of his ethnicity and gender owns a knife, and at that point we're looking at hundreds of suspects who we know could have committed the crime with a perfect copy of the murder weapon because we know one was within walking distance. We know at least two of those knives existed near the murder, so let's spitball that the chance of anyone else using that knife is about equal to the chance of the suspect using the knife. 50-50. It's much more likely that the suspect had his knife than anyone else, but there are so many other people who could have committed the crime that I don't feel comfortable saying he's much more likely than everyone.
(In reality the ownership rate of knives was likely lower among Hispanic teens in 1950's America, the media exaggerated it, but people might borrow murder weapons, so we're into speculation land).
Second, the flaws in the eyewitness testimony are sufficient to basically completely obviate any of the witnesses as interrogators of truth. Eyewitnesses in perfect health and alacrity make major mistakes. There's too much evidence to summarize, but we're talking on the order of 75% chance of correct identification at absolute best. Arguably eyewitness testimony has basically no impact on probability estimates before we start adding in reasons to doubt it. In this context the stated circumstances-that the witness is elderly and saw the murder in the dark through a train-are actually sufficient to make it highly unreliable before we talk about glasses, which are actually just a red herring. Eyewitness testimony is so bad that they should have rejected it before they learned the witness even might be blind! If I was forced to give a prior probably of how likely that eyewitness was likely to pick the right suspect out of a lineup of all suspects in those circumstances, it'd be on the order of one in a thousand. There are likely nearly that many young people of vaguely correct build in a few block radius, and I have nearly zero confidence they could pick one out from another in the stated circumstance. I'm being generous by several orders of magnitude if they have vision problems.
Eyewitness testimony is that awful. It's the source of most miscarriages of justice, and it's arguably simply a mistake to use it in murder trials at all-the correct move is probably to ban it because it's more prejudicial to juries than evidential, even if a glimmer of scientific validity can be gleamed from it by an expert.
Hence the two/three pieces of reasonable evidence are-Knife, and fight with victim (whom was his dad). The fight is not great evidence, but specific threats to family members precede a lot of murders, believe it or not.
The prior probably that kid is guilty because the victim was his dad is around 10% (yes, really), the fight and lack of other familial suspects might increase that to 20%, and as I said I don't consider the knife as raising the odds more than a coin toss. The eyewitness testimony is a rounding error.
I'd hence conservatively put the kids chance of being guilty at ~70%. And that's completely unacceptable for a murder trial. That's barely a preponderance of evidence.
Legally, the correct move at this point is for the jury to walk out and admit everything they've discussed openly, at which point the entire case will be thrown out for a mistrial. Technically Juror 12 could have forced that himself by just saying that he saw an identical knife on the street, which is only a criminal (well...highly legally dubious) act when he smuggles it into the room. That's not good for the boy, because the idea that only one man in twelve was willing to murder him through legal maleficence for being Hispanic in 1950's new York is laughable, but it's the legal option. Juror 12, knowing that, is probably morally justified in pushing for an acquittal anyway.
As for shaming the racist juror into silence-shame is a powerful motivator. Racists, particularly modern racists, are often fundamentally social and physical cowards too, and they just bowled over him (mostly illegally) rather than relying on him to actually admit fault and chance his vote, which is more realistic to the psychology involved. Also fully justified, of course.
We can likely reject your alternative hypothesis because of the prior probably that someone who wanted to rig a trial could do so. There are tens of thousands of people whom might be called for jury duty (accounting for racial and gender discrimination...) on a given day in 1950's New York. Even knowing the day, hundreds are then called, of which 14 get put on a 12 man jury with two substitutes. You can go find the real numbers to do the math if you want, but back of the envelope, without a way to rig that system you're looking at a chance of less than 1/10000 if you knew that the juror wanted to rig the trial ahead of time. Rigging the system to influence juror selection is rare enough to have an infinitesimal prior probability if you know someone wants to. At least as a rogue juror. It's even rare in movie logic.
It's still possible to compromise a jury, but only after it's selected, and that is also extremely rare; if you're sure juror 12 is compromised he's remarkable composed for someone either approached for a massive bribe or whose family is being held hostage, which is how such things were actually done the few times it shows up.