Hacker News Reader

Comment by ceejayoz | Hacker News Reader

Question eight baffles me. Are there characters not used in license plates or something?

>On January 1st, 2018 at 8:00 P.M., John Daly reported a hit and run to the police. Four witnesses were interviewed by the police. They all gave different accounts of the license plate of the vehicle involved in the hit and run. Which of the following plate numbers is most likely to be correct? (A) 381JRZ (B) 358JRZ (C) 351JTZ (D) 351JRZ

I think for a lot of these questions of "four witnesses gave varying accounts" questions, you're supposed to assume that the person who's correct has details most consistent with the majority.

So if you go character by character, everyone agrees the plate started with a 3, 75% agree the second character is 5, etc.

D is the answer that's always consistent with the majority opinion of each character.

Alternatively the police could look up all 4 plates to see which are actually registered, which are on cars that reasonably match other details from the witnesses and forensic evidence from the scene, and talk to the owners of any of the cars that haven't been elimanted. If that fails they could expand the search to other logic permutations of plates. They could even call this process "police work".

This question isn't designed to test how well applicants can master the real world skill of researching license plates. It's designed to see if applicants realize that generally speaking, witness recollections are faulty, but that fact can be mitigated by piecing together a composite based on multiple recollections.

Also, uniformed officers disproportionately come from uniformed officer families. One way they can give their kin a structural advantage is to design the tests to have certain recurring types of logical problems which can be relentlessly practiced. So, year after year, you'll see the same kinds of exam questions, with the details switched around. That license question or a close logical variant may have been on the test for generations.

Right, but looking at those answers there is one plate you'd try first.

I don't know how plates work in the US but in the UK quite often you'll find similar cars with similar plates because they were registered at the dealer in a block. A great example is how about 14 years ago nearly all the police cars in Scotland were silver Ford Focuses with plates that started "SF58" and then had more-or-less sequential runs of three letters. "SF" means it was registered in Edinburgh, "58" means September 2008, and the three letter group is serially assigned. Of course, this applied to all their *unmarked* cars, too...

Plates work very differently on older cars in the US because a car gets a new plate whenever it changes owners or the owner moves from one state to another.

You said the bad word, "work". Sounds like effort... Better to just guess, that's faster and doesn't make your head hurt that much.

This valid for everything, not just police work.

This is reasonable, but the question is then wrong. It should state that the options given all represent plate numbers that were testified to

That’s not how it works for police, they don’t operate under the same standards as a jury for good reason.

Also, It’s generally not illegal to lie to a police officer. You can, and will, be prosecuted for lying to a federal agent.

I think the idea here is that each of the four witnesses gave one of the numbers in the question.

They all agree that it is 3*J*Z.

3 witnesses (all except A) agree that the second digit was 5.

3 witnesses (all except B) agree that the third digit was 1.

3 witnesses (all except C) agree that the second letter was an R.

Every letter/number that D gives is corroborated by two other witnesses (though a different two each time), so D is most likely to be correct.

I think you're right about the intention. However, I think a fairer deduction to draw from this is it proves any individual is quite likely to be wrong so you can't derive anything meaningful.

I think its testing to see if someone can piece together some information based on a majority's reporting, in this case in the individual characters of the license plate. l'm pretty sure the correct choice is D.

I was thinking the same thing, and imagine there’s many standardizations for license plates that the public is probably unaware of.

Nyc taxi's always start with a T and end with a C, for example.

Also, standard issue plates go in allowed sets, eg A123BCD is possible but A1Z3BC0 wouldn't exist.

The four answers are the four responses from the witnesses. Using best-out-of-4 logic for each character the answer is probably (D)

I mean, it seems like the real answer is "take 30 seconds to run all four plate numbers", and then ask the witnesses what color and type of car they saw.

This really puts the training in perspective.

Which plate do you run first?

Doesn't matter. Might be two coincidentally having the same color and model, so it's important to quickly run all of them before the follow-up.

351JRZ. Look at the frequency columnwise.

Guessing it’s the one that has the most people agreeing what the particular characters are.

So the answer is D: 351JRZ?

There is an answer sheet at the end.

Reply on news.ycombinator.com