I replicated an $86M project in 57 lines of code(read.acloud.guru)
read.acloud.guru
I replicated an $86M project in 57 lines of code
https://read.acloud.guru/how-i-replicated-an-86-million-project-in-57-lines-of-code-277031330ee9
11 comments
Having worked for a traffic enforcement company. It is actually more than what meets the eye. ANPR and hardware is probably 5% of the actual infrastructure. Rest of it is generally a pipeline that most of time has to be retrofitted with their existing infrastructure. Lot of the cost into project management and aligning the spec understanding. So when someone says a project is worth $86 million it is not just the hardware and anpr but the whole pipeline integrated with various complicated government workflows, assuring SLO and various other guarantees
This is the 4th time this post came up.
https://news.ycombinator.com/item?id=18681759 https://news.ycombinator.com/item?id=21844164 https://news.ycombinator.com/item?id=20774985 (It was flagged here as a duplicate to 18681759)
Flagging it again.
https://news.ycombinator.com/item?id=18681759 https://news.ycombinator.com/item?id=21844164 https://news.ycombinator.com/item?id=20774985 (It was flagged here as a duplicate to 18681759)
Flagging it again.
I agree that a lot of projects are too expensive but this is a little silly. There is more to a public project besides nailing together a few lines of code. You need documentation, maintenance, user training and lot of other stuff. The actual tech is usually the easy part.
This is like saying I can replicate Google with:
SELECT * FROM websites WHERE title like '%dancing cats%';
SELECT * FROM websites WHERE title like '%dancing cats%';
Dude you've replicated alta vista at best. I'm going to take my google to market and dominate:
SELECT * FROM websites WHERE LOWER(title) LIKE '%dancing cats%';SELECT * from ad_buyers WHERE LOWER(title) LIKE '%dancing cats%';
Medium blog post from 2017, could we add the year to the title please?
[2017] previous discussion https://news.ycombinator.com/item?id=18681759
On the first paragraph I had a feeling I have read that post before. Then I checked the date it was posted and baaaam, 2017.
I'd just like to say something as someone that's sold/installed security cameras and software in both the private, gov and LE sectors.
Yes, there's a lot of "over priced" cameras (and software) out there that are pretty much no better than a $20 usb webcam. That includes the "security" aspect. In my experience (did it for about 3 years), most of these are the white labeled cameras. A majority from China, but Germany and the USA have put out some absolutely dogshit cameras as well. Especially when you factor in the price tag. At least with the Chinese ones, you get a discount for bottom barrel.
At the same time, there's something most people don't really understand about security cameras, longevity. These are like servers (and technically are). They're meant to run 24/7, without stop. Quite a few (if not all newly manufactured one at this point... I hope) can have weekly scheduled reboots, which I always recommend customers set. Most don't and actively refuse to allow a 2 minute block of downtime at an obscure hour. Camera manufacturers know that, and have to deal with it by having a product that just runs constantly.
Next up comes physical protection. You start getting into the $300+ per camera, it's supposed to survive someone hitting it with a baseball bat a few times. Then comes protection from weather and generally being able to survive 120 deg f outdoor conditions or -40 deg f conditions. Your standard camera and embedded board can't do that.
Now, there's a lot of asshole integrators that oversell to governments a lot. My career was primarily based around fixing the last integrator's complete fuck ups. I've seen how government agencies can be bent over and taken for a ride in that industry. Sad part is, none of these folks were even paid off. Just flat out duped. At least the ones I dealt with, I'm sure there's plenty that took money under the table.
On the software side, this is tricky. Yes, the basics of LPR can be done in a few lines of code. But there's more to law enforcement and government accountability, even private sector when developing a case to hand over to the police after an incident. The software company (video management system, VMS), has to guarantee the information has not been tampered. A chain of custody of information has to exist. Video from the VMS cannot be altered in anyway when exported or a lawyer can just wipe his ass with it and throw it out of court. Then there's usability, cross-platform support, driver support for a plethora of devices (there are hundreds of camera manufacturers with hundreds to thousands of SKUs each of devices that one software has to manage). And that's just the tip of the iceberg. I haven't even gotten into general security so it can't be hacked... at least easily (I think Genetec is still top dog in this, it's been 2 years, may have changed. They use to run really good bounty programs, don't know if they still do).
I think 86mil is still retarded as hell because there are plenty of off the shelf solutions that accomplish exactly what they want already, from 2+ years ago. Would only cost them a less than... maybe 8-10 mil in license fees. I don't know how big the city is or needs spec, so it'd take me a while to properly quote... eh, I also never did an international project. I don't know if some of the vendors I dealt with have a different pricing structure for Australia (I know some countries get charged more, for various reasons). So, there's that too.
So just saying, it's a much more complicated issue than most people imagine. The VMS companies do have a large workforce of programmers that have a job I'd never take and do need to be paid. I also over simplified quite a bit and left out other "fun" aspects of selling and deploying physical security... the reasons why I don't do it anymore. Hell, I didn't realize how much of a bitch of a problem it all was until I actually had to do it. I can't blame people for not knowing. Just... be aware, it's not a clear cut situation like this guy makes it out to be.
Yes, there's a lot of "over priced" cameras (and software) out there that are pretty much no better than a $20 usb webcam. That includes the "security" aspect. In my experience (did it for about 3 years), most of these are the white labeled cameras. A majority from China, but Germany and the USA have put out some absolutely dogshit cameras as well. Especially when you factor in the price tag. At least with the Chinese ones, you get a discount for bottom barrel.
At the same time, there's something most people don't really understand about security cameras, longevity. These are like servers (and technically are). They're meant to run 24/7, without stop. Quite a few (if not all newly manufactured one at this point... I hope) can have weekly scheduled reboots, which I always recommend customers set. Most don't and actively refuse to allow a 2 minute block of downtime at an obscure hour. Camera manufacturers know that, and have to deal with it by having a product that just runs constantly.
Next up comes physical protection. You start getting into the $300+ per camera, it's supposed to survive someone hitting it with a baseball bat a few times. Then comes protection from weather and generally being able to survive 120 deg f outdoor conditions or -40 deg f conditions. Your standard camera and embedded board can't do that.
Now, there's a lot of asshole integrators that oversell to governments a lot. My career was primarily based around fixing the last integrator's complete fuck ups. I've seen how government agencies can be bent over and taken for a ride in that industry. Sad part is, none of these folks were even paid off. Just flat out duped. At least the ones I dealt with, I'm sure there's plenty that took money under the table.
On the software side, this is tricky. Yes, the basics of LPR can be done in a few lines of code. But there's more to law enforcement and government accountability, even private sector when developing a case to hand over to the police after an incident. The software company (video management system, VMS), has to guarantee the information has not been tampered. A chain of custody of information has to exist. Video from the VMS cannot be altered in anyway when exported or a lawyer can just wipe his ass with it and throw it out of court. Then there's usability, cross-platform support, driver support for a plethora of devices (there are hundreds of camera manufacturers with hundreds to thousands of SKUs each of devices that one software has to manage). And that's just the tip of the iceberg. I haven't even gotten into general security so it can't be hacked... at least easily (I think Genetec is still top dog in this, it's been 2 years, may have changed. They use to run really good bounty programs, don't know if they still do).
I think 86mil is still retarded as hell because there are plenty of off the shelf solutions that accomplish exactly what they want already, from 2+ years ago. Would only cost them a less than... maybe 8-10 mil in license fees. I don't know how big the city is or needs spec, so it'd take me a while to properly quote... eh, I also never did an international project. I don't know if some of the vendors I dealt with have a different pricing structure for Australia (I know some countries get charged more, for various reasons). So, there's that too.
So just saying, it's a much more complicated issue than most people imagine. The VMS companies do have a large workforce of programmers that have a job I'd never take and do need to be paid. I also over simplified quite a bit and left out other "fun" aspects of selling and deploying physical security... the reasons why I don't do it anymore. Hell, I didn't realize how much of a bitch of a problem it all was until I actually had to do it. I can't blame people for not knowing. Just... be aware, it's not a clear cut situation like this guy makes it out to be.
> To be fair, I have absolutely no clue what the $86M figure includes — nor can I speak to the accuracy of my open source tool with no localized training vs. the pilot BlueNet system.
While I have no doubt that the author's solution is a cleaner approach more suited to real government than the cronyism probably involved in that 86M figure, sensationalizing the difference doesn't help FOSS government advocates. 57 lines of code also doesn't include the libraries involved, so I feel like this is just meant to grab attention. It's pretty rad, but the way it's presented doesn't lend itself to credibility and adoption.