How MTA shut down my app for Penn Station commuters

I can see why the MTA wouldn&#x27;t want you telling people to go to a track before it is actually assigned -- if the train ends up on a different track, now you need to get all those people back up, off that platform an on to the new one, clogging stairs that that people who actually wanted that platform might be trying to use too. Obviously it would be nice to assign tracks earlier, so that people can head straight to the right platform, but sending people to potentially the wrong platform seems even worse.<p>EDIT: Penn is extremely platform&#x2F;track constrained -- NJ Transit, LIRR and Amtrak are all sharing a fixed number of platforms, some of which are too short.<p>To maximize platform utilization, they have to wait until the last minute to finalize track assignments -- if you reserve one too early and the train ends up late, you&#x27;re wasting an empty platform. Once you send a horde of people to a platform, moving them to a different one is a challenge (stairs&#x2F;bottlenecks, communication, etc).

For about a year a decade and a half ago, I commuted into and out of Penn Station with my father who had been doing so since the late 70s.<p>Not only did he know which track any of a half dozen trains that he might take would come in on but he knew where the doors would open for each of the tracks. And he wasn&#x27;t the only one either. If you went down before they announced the tracks you&#x27;d see little clusters of people waiting apart from each other on an otherwise empty track. We&#x27;d usually go to the same car in order to reduce the distance on the other side. Other commuters would do likewise and so trains would have a contingent of regulars.<p>There&#x27;s a fascinating kind of micro-expertise that develops when you do the same thing over and over again.

As an interesting, although unrelated aside: platform data is available in the UK&#x27;s National Rail&#x27;s feeds, but the terms and conditions[1] explicitly prohibit displaying platform numbers early, as mentioned in their developer guidelines[2].<p>I guess the wording doesn&#x27;t technically ban you from displaying historical platform information, but that would likely be a bad-faith use of the data anyway...<p>[1] <a href="http:&#x2F;&#x2F;www.nationalrail.co.uk&#x2F;static&#x2F;documents&#x2F;Terms_and_Conditions.pdf" rel="nofollow">http:&#x2F;&#x2F;www.nationalrail.co.uk&#x2F;static&#x2F;documents&#x2F;Terms_and_Con...</a><p>[2] <a href="http:&#x2F;&#x2F;www.nationalrail.co.uk&#x2F;static&#x2F;documents&#x2F;Developer_Guidelines.pdf" rel="nofollow">http:&#x2F;&#x2F;www.nationalrail.co.uk&#x2F;static&#x2F;documents&#x2F;Developer_Gui...</a> &quot;Occasionally Time-Bound Data will become available through NRE feeds before it is ready to be published to the public. [...] One such example of Time-Bound Data is platform numbers. Early display of platform numbers, particularly at origin and destination stations, can lead to platform overcrowding and&#x2F;or staff not having sufficient time to prepare the train for oncoming passengers. In some instances, platform numbers will be available in Darwin before being displayed on screens in stations.&quot;

Would have been interesting to see if you could reliably crowd-source the data from commuters themselves.

For Penn Station in particular, one can beat the crowd by heading downstairs and looking at the small CRT under the stairs. You will see the platform assignment at the same time as the crowd, but you are closer to the platforms and don&#x27;t have to use the same stairs as everyone else.

There is a similar issue with the MBTA commuter rail in Boston. At south station, the signage in the main waiting area and at the platform only announce the train 10 minutes before departure.<p>However, there are a couple of tools to get around this. For Starters, each revenue service train has a number (like 508). In a push-pull system like the MBTA the first coach in the consist is nearest to to the platform. This coach has a number like 1827. Luckily, the MBTA publishes which trainsets will be assigned to which departures. This makes it easy to know that 1827 is for train 508. All you need to do is walk out to the platform and see if 1827 is sitting there.. if so it&#x27;s your train. You can get this mapping of trainset to train via a bunch of apps.<p>This has been further expanded upon by micro-social apps like &quot;MBTA Rail Tracker&quot; which has a comment section for every train. The whole thread is basically &quot;which platform&quot; followed by a bunch of responses and then snarkiness on why the trains are horribly late all the time.

I&#x27;ve been collecting Penn Station NJT and LIRR track assignment data for almost a year now. I&#x27;ve been intending to use ML to predict track assignment, but haven&#x27;t gotten there yet. I should post it to Kaggle or somewhere similar.

Interestingly enough, I have had this same exact idea for an app, also being one of those commuters who rely on LIRR trains, also being someone who noticed that my peak train home arrives almost always on the same track, and also being one of the (few) commuters who go down to the platform before the train is called.<p>It is a shame that this had to happen to OP. I personally don&#x27;t see the app as an issue, because the amount of commuters that would actually use such a thing is rather low (seriously, stand in the concourse and look around at people, a large amount of them are not using their phones).<p>My original idea was to scrape the data from their webpage, or see if there was a way to get the data from the Train Time app&#x27;s Arrival Countdown page, but according to the post, it has been removed from their website, so there goes that idea.<p>The lesson to me is clear though: don&#x27;t try to make an app that would make commuter&#x27;s lives easier. The MTA does not seem to want that, especially during the track repairs. I find it interesting that it was shut down so close to the start of the Penn Station track repairs...

Why didn&#x27;t he switch to a crowd-sourced data model like GasBuddy?

I wonder if there&#x27;s any correlation or way to know based on where a train is arriving from to know where it&#x27;s headed. If that were the case, you could potentially either look for that data, or if it isn&#x27;t available, have a Waze-like app where commuters can report which train they are on and which track they are pulling into.<p>As an aside, I wonder whether the OP considered charging a hefty price (&gt;$20 or so) for the app. This would lessen the number of users and page requests.<p>Having travelled from Penn many times and having had to deal with the massive stampede that ensues when they suddenly reveal the track number, I would pay dearly to have this info available to me. Even if I used it once every year, not having to fight crowds or look forever to find an available seat, that would be money well spent. Additionally, many of the peak travelers are business travelers who probably have more money to spend on an app. I realize this isn&#x27;t the most democratic solution, but it could be a way to lessen the crowds nonetheless.

It&#x27;s a great explanation of the need for this app, and how zavulon fixed a problem.<p>But organizations change how they work with customers&#x2F;the public all the time, and I couldn&#x27;t help but say Waaah.<p>Things change; tell us about the alternatives you tried to move around this obstacle.

Ok, I understand that the Railway Authority don&#x27;t want people to guess &#x2F; predict platform numbers because of various reasons (problems with platform maintenance, safety and last min changes etc.<p>But what I don&#x27;t understand is that why can&#x27;t they change the seat allocation process such that people can reserve remaining available seats on the trains for free if they have arrived in station lobby? This can result is reduced rush to grab seats and will help people to make a more informed decision around whether to wait for the next train.

Crowdsourcing the track data based on where commuters&#x27; phones travel, something like waze for LIRR should help.<p>Instead of asking users to report the track info, it can be automated to a great extent with the use of BTLE beacons on each platform (but that again needs permission from MTA &#x2F; Penn station authorities unless a long lasting BT 4.0 beacon can be sneaked in somewhere).<p>Sad that MTA is not cooperating. Can understand why the app had to be shut down!

Why don&#x27;t you sue them? It&#x27;s a government agency, and I believe that the data should be available under New York&#x27;s public records laws (I only have limited experience with NY&#x27;s laws and IANAL). From personal experience, being nice to a government agency never got me anywhere, but our lawyer has a 95% win rate against them.

Why would you want to know the train more than 10 minutes before departure? Surely you&#x27;d just get on the immediate next train?<p>Are tickets only for a specific time of train?<p>Or are trains less frequent than every 10 minutes?

www.njtwizard.com<p>We do this for NJ Transit with about 90% accuracy. We spent at least two months figuring out how it all works, though. It&#x27;s non-trivial to do right.

Whoever decided to close his access to the API should be fired.

Why the OP did not continue with scraping the website then? I think there will be no legal issue if he does that.

Scraping the MTA site against the TOS, and intentionally obfuscating requests to limit the chance of the MTA identifying and shutting it down? Yeah, no surprises here.<p>EDIT: Oops, missed that he didn&#x27;t end up going live with the scraping.<p>It would still be great for the MTA to publish this data intentionally; but I&#x27;m sure the potential for trains to switch tracks and the associated backlash when they do is what prevents them from doing it.<p>Perhaps they could adopt the airline model, where there&#x27;s a best effort to reach a particular pre-announced track, with notifications ringing out when they can&#x27;t.