The requirements were surely to give preference to closer delivery people, and the easiest code to write will simply order delivery people by distance.

The simplest solution is therefore that, yes, someone 10 feet away will get it before someone 20 feet away.

The developers would have to go out of their way to add a buffer zone, e.g. find the closest person, add 1,000 feet, find everyone within that range, then select randomly. Or create a "local zone" distance range and a queue. Or other options -- once you try to figure out "what's fair" it gets really complicated.

That extra code is only going to get written if it has to.

So there's nothing ridiculous about it at all. In fact it's the simplest, most expected solution.

Now that it's news, however, Amazon might actually come up with more sophisticated requirements.

This sort of "unintended consequence of over precision" has happened a few times... I remember a while back a dating app (I think?) was returning the distance between the two people with a ridiculous amount of precision, which allowed people to triangulate where someone was with three phones

Here’s an article about how Tinder fixed it: https://robertheaton.com/2018/07/09/how-tinder-keeps-your-lo...

Oh man, I forgot they originally sent the actual location. I still can’t believe how many developers think you can trust the client.

Someone even made a triangulation app to pinpoint other users of Grindr.

Aye, with Tinder, but no extra-devices or actual travel was needed: https://blog.includesecurity.com/2014/02/how-i-was-able-to-t...

Oh right, they just used the API.

Is "cutting the queue" (what this is) an unintended consequence of having a queue? It's perfectly normal behavior. Maybe it gets resolved socially, or the solution could involve tracking whether the follow-up movement happened as assumed (the real hidden assumption). Making a buffer and reducing precision doesn't solve anything actually, it makes it socially more difficult, but it's really a hack, as was choosing the closest physical phone whether a warm body was attached or not.

I think if they randomly select someone with a range, it would make it not as important to try to be physically closer. It wouldn't stop all gaming of the system, but it would be a start.

You'd want to tweak it so you were picking the person in range who went the longest without a gig rather than a naive random pick. Lot more encouraging for the workers vs watching the guy who just got a gig receive yet another before them.

why not randomize selection of everyone within a certain radius rather than just "closest" ? That seems like it would be more fair.

Why would you ever choose to penalize the actual closest driver (gamed algorithm aside)? It seems like a completely reasonable algorithm. In theory, the algorithm sends the closest driver away on a delivery, moving the next closest to the front of an implicit queue.

It would be nice to have perfect knowledge beforehand of how an otherwise reasonable algorithm will be gamed once it's out in the wild.

If there's an algorithm in charge of optimizing, it will be gamed. If there's a single algorithm controlling a large population, it'll be catastrophically gamed.

There is a more general rule than that, I would tie that to Goodhart's law/adage "when a measure becomes a target, it ceases to be a good measure".

It's similar here, when an algorithm is in charge of optimisation it has its heuristics, they will be gamed as they are measures that became a target.

This is such an obvious way to game the system though, if they had put literally any effort into trying to make a system that can't be abused, they would have caught it. Either the system wasn't thought through, or it was thought through by someone with no understanding of how groups of people behave.

It's obvious that people will hang phones from trees?

The reason you're reading this on hn today is specifically because it is novel group behavior.

Na he just means when developing something the first thing you try to do is think how can this be abused.

By handing out jobs to the closest device and not checking the last job that device received is poor programming.

They are probably also measuring by bird distance...

I don't mean to be "that guy" correcting someone's English, but you don't seem to be a native speaker, so I hope this will be well received.

The actual phrase should be "as the crow flies." If you say "bird distance," you might get confused looks from people.

Just trying to help. :-)

I like "bird distance". I knew immediately what was meant, and it was more succinct than "as the crow flies"

I agree, I move that we change the idiom and that pmiller2's remarks be stricken from the record.

Yeah, me too. I didn’t even look twice. And I’m not sure I agree with the advice to use a tired expression rather than something more original.

In Swedish we say ”the bird way”, as in ”it’s x km going the bird way”.

In German we say "the airline" airline is already taken in English.

Air-line, I like it.

In Croatian we use "air distance"... I'm going to be smug about it's more understandable than both (even though I like the Game Of Thrones association when ever I hear "as the crow flies")

But how to know if the bird is flying or driving?

And is it African or European?

Laden or unladen?

No, he said it was a crow, not a swallow

Or if it gets lost along the way. And who says they actually fly in straight lines?

Or swimming :)

The correct term for this measure of the geodesic is orthodromic distance and is guaranteed to get confused looks from people, which is often my intention.

You may then proceed to explain this refers to the distance by camel.

Telling people the shortest path is a great circle is also fun.

Me: 'The shortest distance is a straight line'

You: Except on a globe, it is a great circle.

Me: 'Nobody said on the surface. I'm tunneling through.'

It's more like that on the surface of a sphere, great circles are lines. That's how it's interpreted in spherical geometry.

You're right to gently correct here.

That said, "bird distance" is going in my lexicon.

I read this more as Charlie Day joining hn and I think I prefer it that way.

Honestly, I prefer "bird distance".

I wish this entire thread of replies was an FP of its own. “Slang terms for linear distance through history” (arxiv.org) or something :)

In German, it's "Luftlinie", which literally means "air line" (while airline is "Fluglinie" ("flight line") or "Fluggesellschaft" ("flight company")).

most likely from French "distance à vol d'oiseau" (or "distance d'oiseau" in short)

I'm a french speaker so I also got the reference right away, didn't realize this was not as common in English

I assume the exact details of "bird distance" is found within "bird law".

No, the correct term is “geodesic”.

Bird distance. Not just crows fly straight.

as long as it's not these birds. they'll never get there. https://www.youtube.com/watch?v=V4f_1_r80RY

Don't want to be that guy - then don't.

... and probably only from a specific point. You might be inside the place right at the kitchen's door and be further away in their system than if you stood outside somewhere close to the coordinates of the point.