TSA spent $47,000 on an app that just randomly picks lanes for passengers

TSA discontinued the app last year.

Read the whole story

 

[Mcorther]

Spoiler

 

[Pluvia Arenae]

I would have done it at a cost of 25 cents per pair of lanes (could be less, but the delivered equipment might be too light to reliably handle without frequently getting dropped under the table).

Edit: On second thought, it's a government contract, so after shipping costs, training costs, and a decent profit margin, that should probably be several hundred dollars per lane pair. I guess the TSA got a deal on this one.

 

[jkayca]

Well obviously most of the money went to finance alien research at Area 51

 

[AesirValkyr]

Head down to the comic book shop and buy a D20. Be careful though, if you crit, you must acquit.

 

[SoManyEthicals]

You don't think they spend $40,000 on a hammer, $30,000 on a toilet seat, do ya?

 

[ZaphodHarkonnen]

Sounds pretty cheap actually. If it include device costs, overhead, reviews, the paperwork required to monitor government contracts, etc.

These sorts of things add up really quickly, especially if you're going out to contract and not maintaining an internal resource capable of doing this. Which has its own running costs spread over years.

 

[jimCA]

47k seems extreme, but "a developer could build it in a day" is focusing on the app itself and not anything else.

1) It's a government gig, they didn't need just a dev. There was likely a PM, a tester or two. Probably an analyst. There's inevitably documentation by the ream that goes into anything government.
2) Great, they built it. Where is it installed? How is it installed? So you gotta build that too. And document all of that.
3) It was originally 1.4mln? And then whittled down to 47k? So how much of that was a business analyst being paid to do that work to whittle down the requirements? those meetings add up fast.
4) Was this also that lovely brand of government work where they had to travel to a secure facility to do the work? What travel costs were in the 47k?

 

[Rdrr]

"Many are questioning why a government agency overpaid for the app."

Because it's the government.

 

[Garth]

Probably all the IAP of gems that drove the price up.

 

[stabs]

I been wondering that since they deployed it. I wonder if the iPad cost was part of that figure though.

 

[Faanchou]

ZaphodHarkonnen[/url]":3ewdu0lm]Sounds pretty cheap actually. If it include device costs, overhead, reviews, the paperwork required to monitor government contracts, etc.

These sorts of things add up really quickly, especially if you're going out to contract and not maintaining an internal resource capable of doing this. Which has its own running costs spread over years.

Guess what the missing $1.35 M was spent on.

 

[eric123]

stabs[/url]":zhjytga4]I been wondering that since they deployed it. I wonder if the iPad cost was part of that figure though.

source says it may. no one seems to know.

 

[jdale]

sqlrob[/url]":949grhsv]Randomness is hard, so an app, with all the associated code review and testing, is not as cheap as you might think.

If they didn't already have phones / mobile devices, the physical solutions would probably be cheaper.

This particular function doesn't require true randomness. Pseudo random is good enough, we're not talking encryption here.

 

[briandef]

The best spin I can put on this from the TSA point of view: Like any business, they didn't just ask for a piece of software and massively overpay for just the software, they paid for the software, paid for meetings with devs to actually work on requirements and paid for a support contract for future problems and/or changes.

So that $47k would've paid for future updates to the app (it's not that hard to think of more features the TSA would like it to have) too. But they decided not to continue with the project. It might be $47k in sunk costs, but continuing to spend resources on it would've cost *more* at this point.

Still overpriced, but I think "a newbie could have done it!" isn't quite the right conclusion here.

</best spin>

 

[ZhanMing057]

jdale[/url]":2ouar8e6]

sqlrob[/url]":2ouar8e6]Randomness is hard, so an app, with all the associated code review and testing, is not as cheap as you might think.

If they didn't already have phones / mobile devices, the physical solutions would probably be cheaper.

This particular function doesn't require true randomness. Pseudo random is good enough, we're not talking encryption here.

But think of the one-assignment bias per every 2.7x10^13 passengers!

 

[xpda]

In 2011, Customs and Immigration (also part of Homeland Security) were unable to operate a random lottery. They reneged after notifying thousands of foreigners that they can immigrate to the U.S., saying the random results were not random. I wonder how much they spent on that.

http://spectrum.ieee.org/riskfactor/com ... the-glitch

 

[MacCruiskeen]

eric123[/url]":2j0qtxyl]

stabs[/url]":2j0qtxyl]I been wondering that since they deployed it. I wonder if the iPad cost was part of that figure though.

source says it may. no one seems to know.

Yeah, the released information isn't really informative about how the money was spent or what was really delivered for that money. It's just a number.

 

[MahNerd]

An app that they spent 47,000 dollars on then discontinued, a website that cost 800 million by the time they managed to get it in decent shape.

I don't know about any of yall but I'm sensing a pattern here.

 

[Skelator123]

briandef[/url]":2uh8rnc9]The best spin I can put on this from the TSA point of view: Like any business, they didn't just ask for a piece of software and massively overpay for just the software, they paid for the software, paid for meetings with devs to actually work on requirements and paid for a support contract for future problems and/or changes.

So that $47k would've paid for future updates to the app (it's not that hard to think of more features the TSA would like it to have) too. But they decided not to continue with the project. It might be $47k in sunk costs, but continuing to spend resources on it would've cost *more* at this point.

Still overpriced, but I think "a newbie could have done it!" isn't quite the right conclusion here.

</best spin>

Yup. Any large business or government operation is going to spend around that as a baseline for any software. They probably spent more on operational readiness testing than actual development.

It's actually not all that bad, and I'm far from a TSA defender or apologist.

 

[briandef]

jdale[/url]":1vhaplx1]

sqlrob[/url]":1vhaplx1]Randomness is hard, so an app, with all the associated code review and testing, is not as cheap as you might think.

If they didn't already have phones / mobile devices, the physical solutions would probably be cheaper.

This particular function doesn't require true randomness. Pseudo random is good enough, we're not talking encryption here.

Encryption is definitely harder to get right, but coders screw up pseudo random numbers too, more frequently than I like to think about.

http://www.tedunangst.com/flak/post/random-in-the-wild is a fun read.

 

[primordius]

ZaphodHarkonnen[/url]":15mbtpcg]Sounds pretty cheap actually. If it include device costs, overhead, reviews, the paperwork required to monitor government contracts, etc.

These sorts of things add up really quickly, especially if you're going out to contract and not maintaining an internal resource capable of doing this. Which has its own running costs spread over years.

Agreed.
Software development has hidden costs that most people don't understand.

Furthermore, it is very likely that this app started off as one thing, and due to changes or other fingers in the pie, it devolved/evolved into what it ended up being... a random assignment.

 

[CatOne41]

This is the perfect opportunity for a good hash algorithm.

- Passengers get O(1) distribution to a line of equal length to all others
- Screeners really get to kick back during their break. Dude.

 

[ARS_dabbler]

Just flip a coin. Two coins for 3 - 4 lanes. Three coins will handle up to 8 lanes. Plenty of coins collected by the TSA every day left behind by hurried passengers.

 

[Old_Fogie_Late_Bloomer]

My heart wants to believe that they built an app based on thousands upon thousands of requirements, A/B'd it in testing with an app that flipped a virtual coin, and found that the coin performed slightly better than the specs. :/

 

[LeftCoastRusty]

So...they needed an app, and went app shopping. I;m guessing the pricing conversation went like this:

TSA: "how much will that fine piece of software cost?"

Vendor: "Let's try out the random number capabilities and see.....Look! It's $47,000!"

 

[vassago]

MahNerd[/url]":1mj50zdt]An app that they spent 47,000 dollars on then discontinued, a website that cost 800 million by the time they managed to get it in decent shape.

I don't know about any of yall but I'm sensing a pattern here.

That the system of contracting based on political ties and lowest bidder is a shit system with shit results and that the government would be best served by actually ponying up for a decent IT and software engineering infrastructure that would likely end up saving them money in the long run and have better results?

 

[Comradephate]

sqlrob[/url]":14g39hek]

jdale[/url]":14g39hek]

sqlrob[/url]":14g39hek]Randomness is hard, so an app, with all the associated code review and testing, is not as cheap as you might think.

If they didn't already have phones / mobile devices, the physical solutions would probably be cheaper.

This particular function doesn't require true randomness. Pseudo random is good enough, we're not talking encryption here.

Yeah, it does require true randomness. If someone else watching can predict, you just opened a hole in your security. rand(), quite frankly, sucks. Maybe it's good enough for this, but that's testing, which, guess what, costs money.

It's honestly baffling that you have a name like "sqlrob" but you don't believe that there are any modern implementations of random number generators where the outcome can't be predicted by even a fairly interested observer in a short period of time.

Even something as absolutely trivial as generating a seed from the device's accelerometer position on app startup (and then regenerating every period of time) and then using the current millisecond or whatever to generate "randomness" would be too difficult to predict given how small the sample size would be. Even if you get hundreds of samples before it resets, and you actually know what happens at the end of each line (you wouldn't) you still won't be able to reliably place yourself in the correct line.

You'd have a much better shot if you just get in line and then after a while act like you have to take a shit, come back, try again. Get a fake doctor's note that says you have IBS or something. You'd probably end up in the line you wanted before the TSA got suspicious of you.

 

[gmerrick]

So in all of this, there really is an app for just about everything!

 

[egoebelbecker]

In all fairness there was a $23,500 version, but it required IE7, Flash, and you had to allow pop-ups.

 

[Peevester]

AesirValkyr[/url]":26dawhyd]Head down to the comic book shop and buy a D20. Be careful though, if you crit, you must acquit.

We already know what happens if you fumble, you end up with the TSA.

Seriously guys, buy some old games of Sorry off of ebay and use the dice popper. Thanks, that will be $46,000 please.

 

[snowman<ca>]

It is like they purposely try to be the most annoying federal agency

 

[Carpetsmoker]

Well, some schmuck paid $1 billion for Instagram. Overpriced apps seem the default, rather than the exception. All things considered, this is a fairly mild example.

 

[snowman<ca>]

vassago[/url]":2z6j8zr7]

MahNerd[/url]":2z6j8zr7]An app that they spent 47,000 dollars on then discontinued, a website that cost 800 million by the time they managed to get it in decent shape.

I don't know about any of yall but I'm sensing a pattern here.

That the system of contracting based on political ties and lowest bidder is a shit system with shit results and that the government would be best served by actually ponying up for a decent IT and software engineering infrastructure that would likely end up saving them money in the long run and have better results?

If you think the bidding system gets bad results, wait till you see what happens when a department inside an agency gets a big enough budget they can run their own shadow IT projects.

 

[D-Fresh]

Joke's on the taxpayer, the TSA is actually just rocking a revamped I Am Rich app

 

[NetworkElf]

If I didn't work government, I wouldn't believe this.

However...

 

[Rabid Badger]

I just found a dozen free random number generator apps on Google Play with ratings of 4 stars or higher in under 60 seconds.

 

[Akemi]

Faanchou[/url]":1pb8wvn6]

ZaphodHarkonnen[/url]":1pb8wvn6]Sounds pretty cheap actually. If it include device costs, overhead, reviews, the paperwork required to monitor government contracts, etc.

These sorts of things add up really quickly, especially if you're going out to contract and not maintaining an internal resource capable of doing this. Which has its own running costs spread over years.

Guess what the missing $1.35 M was spent on.

Shaved Brazilian hookers and a Tony Montana sized mountain of blow? Wait, my bad. That's the secret service.