I would be hella surprised if the $86M solution was better than what this guy would make given $8.6M. So there's not a lot to debate around the overrun costs of government purchasing.
Plenty of ethics to debate, but when it comes down to being wasteful with tax dollars, there's a monopoly on that.
>Plenty of ethics to debate, but when it comes down to being wasteful with tax dollars, there's a monopoly on that.
Is it the government that's being wasteful with tax dollars, or the company that took the money? Why is there this idea that business can be as wasteful as they want, but it's the governments fault?
Every corporation, every private charity, and every public institution that I've worked for has been wasteful, and the corporations have been the worse offenders in that regard.
I don't doubt _this_ guy (and most HN-ers) could do a great job for $8.6M, but there are plenty of companies that couldn't. Govt contracts cost so much in part due to the number of legal hoops you have to jump through to get one. It's better to get something that works and is mostly supported even if it costs a lot than to get nothing for still a lot.
> It's better to get something that works and is mostly supported
But a lot of companies fail at even this basic test. It doesn't work, and it's unsupported. Meanwhile, they're getting paid more to 'fix' it while making the users run around in circles.
Sure, but imagine how much worse things would be if the bar were lower. Govt contracts always bring out the leeches - it's not unique to software contracts.
Part of getting older is accepting things like this as being how we stop truly shitty jobs from getting done.
How would a non-technically competent government official figure out which random hacker to award a contract like this to? They can't. And if they tried they'd probably hire someone that would over-promise and under-deliver and wouldn't be around in six months to fix things if they needed fixing.
I've no doubt there was a chance to save some money, but saying this guy replicated an $86M system is quite a stretch. He managed to replicate the code to OCR the plates in a somewhat controlled condition, and use existing networks to access an existing web service. It's not going to work in low light conditions traveling at 55MPH with plates at varying distances while simultaneously fighting off attacks to jam/DOS/insert false results during a power outage. Additionally there's going to be a backend to store and process the data for whatever purposes. We really don't know much about what the scope of the $86M project was, and may have included a lot of unrelated things also.
> So, although I jokingly named the project plate-snitch when I set it up on my computer, I’m now faced with the conundrum of whether to report what I saw.
> Ultimately, the driver was detected using a prototype of a police-only device. But driving on a 2016 registration (canceled, not expired) is a very deliberate move. Hmm.
For me, the bigger issue is the integrity part of sending each plate to their server. This would in extension allow tracking the location of each car.
I think it would be very interesting with a government founded project with several cheap cameras, each updated daily with a local database of stolen cars. That way the camera only need to notify someone when it actually sees a stolen vehicle. There is of course the risk of this being abused as well, but not as big.
The problem is that once you have the ability to do that sort of surveillance, every agency in your jurisdiction will want to use it for all kinds of other purposes.
Australia is... let's just say.... not a "privacy-focused" nation. Most people here love the Nanny-state, as for any given individual, most of the rights being trampled are things those icky "others" enjoy.
I was at a conference yesterday in the US where they casually mentioned they have a drone that can (now) find devices (phones) from 3 miles away to within 10 feet (or maybe 10 meters?) from a current IP address. The level of surveilence is out of control.
It was a small business conference in Wichita KS where the DOD was encouraging small business to work more with them. One of the speakers used that as an example of a successful project. When the slides come out (hopefully in a few days) I’ll look through them to see if I can find any specific to look up, and come back here and post.
I like that his followup is truthful about the very real short comings of his solutions. For any real-time system, if there is a hard requirement of 95% accuracy, I can understand the initial pricetag. The vendor carries all the risk. Making proof of concept resilient and usable in the real world is not easy nor cheap.
His examples of camera placement and image distortion alone are non trivial problems to solve. Maybe they need custom made cameras for them to be fast enough and small enough, with good enough resolution.
That said, $86M is a lot of money. If you assemble a team of 5 people. Assume they use $1M a year each in for salary, office, equipment, ect, they can work fulltime on this for 5 years and still have more than $60M left for other stuff.
Add a way to compensate for distortion from windscreens, facial recognition and access to the driver's license database, and you can check for those driving with an expired driver's license.
Correlate the car registration details with the driver ID and you have a way to find cars that are potentially stolen ( authorised drivers of the car would need to on a database )
> ...and you have a way to find cars that are potentially stolen...
You’ve never had a spouse borrow your car to go to the store, have you. Also, the cops basically do this now: it’s called “pulling you over and asking for your license and registration”.
Plenty of ethics to debate, but when it comes down to being wasteful with tax dollars, there's a monopoly on that.