Traffic report on notification screen, how i think it miscalculates

[Schtibbie]

So I've noticed something interesting about my phone's traffic-on-your-way-home-from-work predictions on the notification screen. I suspect that Apple wrote it such that it looks at *current* traffic (right now) for the route home on each stretch of that road NOW (as if you'd drive each stretch of your route NOW).

But the thing is, you *don't* drive like that. I get to the nasty stretch of stop-and-go 20 minutes into my commute if I start NOW. At 5:00, that congestion doesn't exist. At 5:20, it *always* exists on that stretch every day Monday - Friday. It makes no sense for the software to look at how that piece of road is faring at 5:00 and use that math. It consistently tells me "Traffic is unusually heavy" (which it isn't - it's ALWAYS this heavy) and that it'll take me 35 minutes. It takes me 50 minutes every time. Not 35. It doesn't learn. And it doesn't look at last Friday's traffic on those roads at the times one would arrive at those points to make a prediction. It could but it doesn't.

Thoughts?

 

[Merkyworks]

In a perfect world I agree with you and your point but I'm not sure if that is very practical. The amount of data used would be much higher if it did past traffic history calculations, also what if there as a major wreck the following week? This would then over estimate the time of this weeks commute. Its not possible for it to predict traffic so when it does calculate time of drive it has to base it off of current traffic conditions. For certain commutes were traffic is very time dependent I would think you would see the estimated time and know you have to add "X" amount to it for a more accurate time. That's all the notification is anyway, an estimate of your commute.

 

[Schtibbie]

In a perfect world I agree with you and your point but I'm not sure if that is very practical. The amount of data used would be much higher if it did past traffic history calculations...

Yeah, I know it would be nontrivial. But I don't think I'm asking for gold-plating. My point boils down to this: it is invalid for a time estimation to use traffic speeds at each point along a route *now* as though you will exist and travel at those points *now*, all at once. Especially when certain major thoroughfares are universally locked down at 10mph (not their 60mph limit) every single weekday always at 5:20 (for example). If it's 5:00 and I'm miles away from that spot, there's no way it's gonna be 60mph by the time I get there. It's gonna be 10mph. And Apple knows that. Their data knows that.

 

[Tyler23]

Yeah, I know it would be nontrivial. But I don't think I'm asking for gold-plating. My point boils down to this: it is invalid for a time estimation to use traffic speeds at each point along a route *now* as though you will exist and travel at those points *now*, all at once. Especially when certain major thoroughfares are universally locked down at 10mph (not their 60mph limit) every single weekday always at 5:20 (for example). If it's 5:00 and I'm miles away from that spot, there's no way it's gonna be 60mph by the time I get there. It's gonna be 10mph. And Apple knows that. Their data knows that.

It's live traffic data, not predictive. It doesn't want to guess about the future, even if it's the case 99% of the time.

If you know how long it's going to take you to get home and when there will be traffic, why even bother with it?

For me, I commute against traffic, so my normal commute is 27 minutes. Occasionally, there will be rain or an accident either going to or from work, and it will cause a backup. When this happens, my traffic info in NC automatically updates and tells me that there is unusually heavy traffic, so I know to be ready for it and can plan accordingly.

 

[Max(IT)]

That's why I disable it.
Useless drain of battery. It's not predictive, so on a short trip it could work, but there are common situations (like the one you described) where it is very unreliable.

 

[hafr]

I half-agree, although I have no idea how difficult it would be to implement. A system that looks at historical data and give hints such as "there are usually long queues from around 5pm, so maybe you want to wait until after 6:30pm if possible".

Although this wouldn't be the least bit useful for commuters since they know it already, and those travelling new routes who don't know it won't see the information in NC...

 

[Max(IT)]

I half-agree, although I have no idea how difficult it would be to implement. A system that looks at historical data and give hints such as "there are usually long queues from around 5pm, so maybe you want to wait until after 6:30pm if possible".

Although this wouldn't be the least bit useful for commuters since they know it already, and those travelling new routes who don't know it won't see the information in NC...

How difficult ? On a technical base it could be easy, but dude there are a lot of places in the world to take account of

 

[SusanK]

Yesterday a One to One trainer was chatting with me for a considerable period of time. He was talking about this function with effusive praise. I just smiled and listened. Had no clue on why this helped him to and from work.

I've been successfully getting myself from home to anywhere and back before smartphones.

This function sounds ( to me) as more non useful fluff Apple is cluttering up the iOS with.

A little substance would go a long way.

It was very telling that staffers has do much time to visit with me. Store not busy. No lines. No hint of launch day. As they say on the tv shopping channels " there're flying out of here"

 

[BrianBaughn]

Estimated travel times will probably become more and more accurate as time goes on and more data is gathered and analyzed.

 

[SusanK]

iWork too

Forgot to mention the trainer was going on and on about the new iWork too. He said he was working with a gentleman who spent the last 10 years in a country absent technology. He mentioned the country but I was zoning out by then.

Matt (trainer) said the man was amazed with iWork especially Pages. They have spent hours on session for Pages.

If this imaginary One to One client is amazed by new Pages he must have been in a cave for a long time. I have an old Mac running Claris Works 3.0. More functionality on the old Claris than the new Pages.

I have Pages on my iPad. Never considered or expected it to be a fully functional word processor. I use it for grocery lists and things like that.

Goes without saying Matt lost all credibility with me yesterday. Nice guy, but really!

 

[Cscottrun]

So I've noticed something interesting about my phone's traffic-on-your-way-home-from-work predictions on the notification screen. I suspect that Apple wrote it such that it looks at *current* traffic (right now) for the route home on each stretch of that road NOW (as if you'd drive each stretch of your route NOW).

But the thing is, you *don't* drive like that. I get to the nasty stretch of stop-and-go 20 minutes into my commute if I start NOW. At 5:00, that congestion doesn't exist. At 5:20, it *always* exists on that stretch every day Monday - Friday. It makes no sense for the software to look at how that piece of road is faring at 5:00 and use that math. It consistently tells me "Traffic is unusually heavy" (which it isn't - it's ALWAYS this heavy) and that it'll take me 35 minutes. It takes me 50 minutes every time. Not 35. It doesn't learn. And it doesn't look at last Friday's traffic on those roads at the times one would arrive at those points to make a prediction. It could but it doesn't.

Thoughts?

If it predicts traffic, there is the possibility that it will tell you traffic exists, when it doesn't. Why would you want that?

While I understand some roads (I live in the tri-state area, I know best), are always filled with traffic during rush hour; what about the roads that are clogged up because of construction for a day or two, or a major accident? How could it reliably predict traffic with these variables?

 

[GreyOS]

Yes, it should look into the future so its predictions are what you know them to be after the fact.

 

[Merkyworks]

Clearly the people why want the travel time function to predict traffic have no technical understanding of computer logic or how programming actually works. What you want is not possible, the best it could do is build a database of traffic history (mind you this would require another massive Apple data center to collect traffic worldwide) and then use this data to predict current conditions.

HOWEVER lets just say they did this, the trend history for something like this would be maybe 1 month back for predictive traffic reports. Anything longer would not be relevant enough. So a lot of the time when you have major freeway construction it will last for 2-6 weeks lets say. SO once the construction is finally done and the freeway moves fast the historical data will still be predicting that the traffic is very high and the commute time is long. What then!? This is exactly why the app is meant to give you conditions based off of current local traffic flow.

 

[hafr]

Clearly the people why want the travel time function to predict traffic have no technical understanding of computer logic or how programming actually works. What you want is not possible, the best it could do is build a database of traffic history (mind you this would require another massive Apple data center to collect traffic worldwide) and then use this data to predict current conditions.

HOWEVER lets just say they did this, the trend history for something like this would be maybe 1 month back for predictive traffic reports. Anything longer would not be relevant enough. So a lot of the time when you have major freeway construction it will last for 2-6 weeks lets say. SO once the construction is finally done and the freeway moves fast the historical data will still be predicting that the traffic is very high and the commute time is long. What then!? This is exactly why the app is meant to give you conditions based off of current local traffic flow.

You say I have no technical understanding of computer logic or how programming work, so maybe I'm completely wrong here - but wouldn't it be fairly easy to tell a traffic surveillance program to separate/ignore times according to flag when finding averages?

For instance, before a construction work has started it might say the average travel time will be 30 minutes, from the day that it starts it might say the normal travel time is 30 minutes, but there is construction work going on so expect some disturbances, after a few days it has found a new de facto average speed on that route, and when construction work is done - it just completely ignores whatever information was fed into the system since the construction work was flagged.

I mean, it shouldn't be any harder than ignoring data from weekends or accidents, right?

 

[Tyler23]

You say I have no technical understanding of computer logic or how programming work, so maybe I'm completely wrong here - but wouldn't it be fairly easy to tell a traffic surveillance program to separate/ignore times according to flag when finding averages?

For instance, before a construction work has started it might say the average travel time will be 30 minutes, from the day that it starts it might say the normal travel time is 30 minutes, but there is construction work going on so expect some disturbances, after a few days it has found a new de facto average speed on that route, and when construction work is done - it just completely ignores whatever information was fed into the system since the construction work was flagged.

I mean, it shouldn't be any harder than ignoring data from weekends or accidents, right?

How would the system know the construction work was done and to ignore that data?

As the other poster mentioned, this is just completely unrealistic, would require much more manpower, and is just not feasible nor useful enough to attempt to implement.

 

[TheKrs1]

It was very telling that staffers has do much time to visit with me. Store not busy. No lines. No hint of launch day. As they say on the tv shopping channels " there're flying out of here"

I'm glad you had all that free time to waste so that you could keep that apple store employee busy. Can you please tell that store to transfer him to either of my Hometown Apple stores? I've seen numerous people leave because the wait for an associate is too long. (even in the middle of the workday).

---
I think the biggest flaw with the frequent locations data is that it completely ignores my location information in my Calendar. It's cool to track where you think I'm going to be, if you have no clue... but since I'm telling you where I should be with my Calendar locations.. go ahead and use that data!

 

[Merkyworks]

For instance, before a construction work has started it might say the average travel time will be 30 minutes, from the day that it starts it might say the normal travel time is 30 minutes, but there is construction work going on so expect some disturbances, after a few days it has found a new de facto average speed on that route, and when construction work is done - it just completely ignores whatever information was fed into the system since the construction work was flagged.

This could maybe work except that scheduled construction does not always end/start when it is planned to. What happens then? local city traffic sites do not always have the most up to date info and that is where Apple will get its info. Also how is Apple to know what time multiplier "factor" to add for the increase in traffic? Your wanting a computer program that allows for the user to not use common sense. Computer are supposed to help us become more efficient and smarter, not lazier and dumber.

Only further fuels my prediction that the move "Idiocracy" is going to be reality some day

 

[MozMan68]

It's only there to tell you how long it will take IF you left right then when viewing it.

So, if it takes me 45 minutes to get home normally and I suddenly see it says 60 as I'm leaving, there must be an accident, or traffic, or construction.

Not perfect, but it is nice to verify at a glance.....(and then I use Waze while driving so I have live updates...)

 

[Schtibbie]

It's only there to tell you how long it will take IF you left right then when viewing it.

Yes, I know. But I'm claiming that it isn't correct for it to tell me the traffic on a busy section of highway that's 20 minutes away from me *now*, because I'm not THERE now. I'm HERE, and I'm leaving NOW. In 20 minutes I'll be there. And in 20 minutes (for example), every single day that section of highway goes at 10mph. What the app appears to do is look at how it flows *now* even though it knows it will take me 20 minutes to get to that spot. That's wrong.

This is a solvable problem. And to the poster who claimed that I (or others) don't know about programming or whatever - I AM a programmer. C, prolog, LISP, pascal, Java, Objective-C, SQL, and a wee bit 6502 assembly back in the day. Also Fortran.

 

[Tyler23]

Yes, I know. But I'm claiming that it isn't correct for it to tell me the traffic on a busy section of highway that's 20 minutes away from me *now*, because I'm not THERE now. I'm HERE, and I'm leaving NOW. In 20 minutes I'll be there. And in 20 minutes (for example), every single day that section of highway goes at 10mph. What the app appears to do is look at how it flows *now* even though it knows it will take me 20 minutes to get to that spot. That's wrong.

This is a solvable problem. And to the poster who claimed that I (or others) don't know about programming or whatever - I AM a programmer. C, prolog, LISP, pascal, Java, Objective-C, SQL, and a wee bit 6502 assembly back in the day. Also Fortran.

I'm not sure you fully understand?

It's not telling you how long it will take to get where you already are, which seems to be what you're implying. It's telling you how long it will take to get where you're going. You're saying there's traffic there now, and that it will be there when you get there..So the traffic data in NC is telling you there will be traffic on your way home..

It's telling you how it is now, and that if you leave now, you will encounter that traffic, because it cannot accurately look into the future as to whether or not that traffic will be there, and it would be silly to try.

 

[Schtibbie]

I'm not sure you fully understand?

It's not telling you how long it will take to get where you already are, which seems to be what you're implying

No, my commute takes 50-60 minutes. About 20 minutes in, I hit SLOW traffic (10mph) on a main highway. That highway always goes to 10mph at 5:20pm. At 5:00pm it's 60mph. Awesome, right? Well, it's 5:00 and I'm at my office and getting into my car to leave. What's the speed on highway X? 60mph. When I GET to highway X it'll be 10mph. Apple's traffic data has this info, somewhere in their databases. At least for major highways they do. But the math they appear to use is what speed is highway X RIGHT NOW. I'm claiming that's invalid. I'm still 20 minutes away from highway X.

 

[C DM]

So I've noticed something interesting about my phone's traffic-on-your-way-home-from-work predictions on the notification screen. I suspect that Apple wrote it such that it looks at *current* traffic (right now) for the route home on each stretch of that road NOW (as if you'd drive each stretch of your route NOW).

But the thing is, you *don't* drive like that. I get to the nasty stretch of stop-and-go 20 minutes into my commute if I start NOW. At 5:00, that congestion doesn't exist. At 5:20, it *always* exists on that stretch every day Monday - Friday. It makes no sense for the software to look at how that piece of road is faring at 5:00 and use that math. It consistently tells me "Traffic is unusually heavy" (which it isn't - it's ALWAYS this heavy) and that it'll take me 35 minutes. It takes me 50 minutes every time. Not 35. It doesn't learn. And it doesn't look at last Friday's traffic on those roads at the times one would arrive at those points to make a prediction. It could but it doesn't.

Thoughts?

So is this different than how Google does it, or how any other GPS providers do it? If so, how do they do it--and how do we know that's the case--and is it actually right/better most of the time?

 

[jr866gooner]

Just looked at privacy and went to the history of the frequent locations. It stamped me at work between 8:34 and 15:31 yesterday and 7:51 -16:35 and both not necessarily correct. When/how often does it start to " know" you are In a location? Does it take a few snapshots to determine your still in said location?

Clever that it keeps tabs on you, not sure if I'll remember to use it's full functions.

 

[hafr]

How would the system know the construction work was done and to ignore that data?

As the other poster mentioned, this is just completely unrealistic, would require much more manpower, and is just not feasible nor useful enough to attempt to implement.

The same way normal GPS's know when to stop showing that little construction triangle, perhaps?

 

[hafr]

This could maybe work except that scheduled construction does not always end/start when it is planned to. What happens then? local city traffic sites do not always have the most up to date info and that is where Apple will get its info. Also how is Apple to know what time multiplier "factor" to add for the increase in traffic? Your wanting a computer program that allows for the user to not use common sense. Computer are supposed to help us become more efficient and smarter, not lazier and dumber.

Only further fuels my prediction that the move "Idiocracy" is going to be reality some day

Your assumptions are wrong, this is a strictly theoretical discussion for me. I don't use GPS apps to tell me exact travel times or traffic info.