00:00 - 00:03 | We're looking to model a rail line here |
00:04 - 00:05 | And a toll project here |
00:05 - 00:07 | A multi use path through here |
00:08 - 00:12 | We're upgrading bus to BRT here |
00:12 - 00:15 | We're looking at a new highway here |
00:17 - 00:19 | I see... |
00:19 - 00:21 | The new model will work well with these |
00:24 - 00:26 | Boss... |
00:27 - 00:28 | The new model... |
00:31 - 00:33 | The new model gives the same results |
00:34 - 00:36 | There's no substantial difference in results |
00:53 - 00:58 | All the non-modelers please leave the room. |
01:13 - 01:15 | WHAT THE FUCK DID WE PAY FOR? |
01:15 - 01:17 | THEY TOLD US THIS WOULD BE BETTER |
01:18 - 01:23 | THE NEW MODEL WAS NOT AN ACADEMIC EXERCISE |
01:25 - 01:28 | AND THIS DAMN THING REQUIRES A TON OF DATA |
01:29 - 01:31 | AND IT TAKES FUCKING DAYS TO RUN |
01:31 - 01:34 | THE OLD MODEL COULD RUN IN 8 HOURS |
01:34 - 01:37 | EIGHT FUCKING HOURS!!! |
01:37 - 01:40 | Why the fuck did we do this? |
01:40 - 01:42 | The DOT said it would be better |
01:42 - 01:46 | FUCK THE DOT |
01:46 - 01:48 | So did the consultants |
01:48 - 01:52 | The consultants WROTE the fucking studies! |
01:53 - 01:54 | They want to make money! |
01:56 - 01:57 | The consultant walked away with a shit ton of money |
01:57 - 02:00 | and all we have is a model that gives the same results |
02:00 - 02:03 | And takes four times as long to fucking do it! |
02:04 - 02:08 | And attractions are still as bad as in the old model |
02:08 - 02:13 | And the truck modeling sucks ass too |
02:14 - 02:16 | And all for the same result |
02:17 - 02:21 | If this was an improvement, it would be faster and better |
02:27 - 02:29 | And our RMSE is still 39% overall |
02:30 - 02:34 | Nothing is better |
02:34 - 02:36 | More money, more data, and none of which we have |
02:41 - 02:42 | Now all of our data is stuck in memory |
02:43 - 02:47 | So we can't determine accessibility improvements for EJ groups |
02:48 - 02:53 | Nobody has done any before and after studies |
02:54 - 02:56 | Or sensitivity analyses |
02:56 - 02:59 | The nonmotorized modes suck too |
03:00 - 03:02 | Good thing we don't have to actually model that bike trail! |
03:04 - 03:07 | Don't worry, we're still building that multi-use path! |
03:14 - 03:16 | This is bullshit |
03:19 - 03:23 | Millions of tax dollars spent on this... |
03:25 - 03:26 | Nothing but the same results to show for it |
03:31 - 03:33 | We have to give this thing more data |
03:40 - 03:46 | And more complex data |
03:46 - 03:49 | And we'll have to write a $500,000 check for recalibration in a few years. |
03:53 - 03:56 | On top of our $25,000 model server |