An algorithm wipes clean the criminal pasts of thousands | Hacker News Reader

An algorithm wipes clean the criminal pasts of thousands

bbc.co.uk

152 pointsjjp54 comments

It sounds like Code For America is doing great things! That said, I wish the title wasn’t tech focused, since that’s not the real news here, and the red-herring implication that it might be some kind of algorithmic mistake is not just completely irrelevant, but subtly pushing the narrative that tech is out of control and fallible — the exact opposite of what happened here.

Possibly the most important piece of summarizing information in this article is that a person in the position to do so actually decided to expunge records automatically, without waiting for those convicted to defend themselves. That’s pretty good news!

Perhaps the word “helps” should be added to the headline to indicate that this is working as intended, and there’s a human in loop. It could be more clear that the tool is simply overcoming cumbersome process.

Seems almost intentional on BBC's part.

Hey, I'm on the team at Code for America that built these tools. I really appreciate your comment because you're exactly right - the issue is and has always been one of policy and implementation. One of the core parts of our mission at CfA is to "show what's possible." In this case that means showing DAs and legislators that proactive record clearance is possible, simple, and good for communities.

As other commenters are pointing out what we built amounts to a few "if" statements and an excel macro. They're right! That's a good thing too, because it means that governments are well positioned to follow our work without needing a high-cost technology partner or RFP.

If you're in California and would like to ask your state reps and senators to support more legislation what would require proactive record clearance please take a look at AB 1076 and AB 972 (linked below).

AB 1076: https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml... AB 972: https://leginfo.legislature.ca.gov/faces/billTextClient.xhtm...

I love the point that the simplicity of implementation is one of the strengths here, that's true!

Does Code For America have room for occasional and/or part time contributors... people with full time jobs that would love to help for a few hours a week here and there?

You should totally check out our brigade network: https://brigade.codeforamerica.org/

While they brigades aren't always working on the same projects as full time staff they're often doing high impact work and are super well connected to their local communities. They've built some really great projects in partnership with their local governments and CBOs.

I'm a member of the KC brigade and I know there is a team working on this for MO. Similar law was passed.

> "As part of the new law, those with prior convictions could now seek to have them struck off their record."

I really don't get the thought process that leads to this kind of thing. "Let's make it so we have to have to implement a new system that people petition and we review to determine if we expunge their record."

Versus: "The new standard is if the conviction was for X, and the person does not have Y on their record, (etc.), then the record of X conviction is considered expunged. And then behind the scenes automatically make it so."

Not sure a record can be expunged without an order from a court targeting a specific conviction/record in the separated system in the US.

Here in Turkey, we just disregard government records older than a certain age (80 years past crime for criminal records for instance). If a record is completely disregarded in proceedings, it could be separated into archive or expunged safely. It is just one more day job for your faithful computer.

That's not the case in the US. The only real automatic sealing is that the criminal records of kids are generally sealed.

One of the reasons they can do that in Turkey is Turkish government is structured in such a way that most records of even local government organisations are required to be entered into central database. USA lacks that.

I could be wrong, but I believe the logic behind this is that the law is still paramount and because the person did in fact break the law (in this case a felony crime), they should have to apply for their conviction to be expunged and make the case in court. Because at the end of the day, while we might not consider what they did a big deal now, when they committed the act it was a serious crime.

I also think they want manual reviews because there could be plenty of cases where they only charged the person with possession for personal use but might actually suspect them of worse crimes like intent to distribute or something else like possible gang affiliation. So I don't think they ever really want there to just be a blanket "oh there's no charge of a violent crime so we can wipe it" because they might be wiping the record of a known gang member or serious criminal but didn't have quite enough evidence to charge and convict them on the other offenses.

So basically, don't expunge records for people who are innocent of other crimes, but who we think are bad people?

How is this not a punishment outside of due process?

> but might actually suspect them of worse crimes like intent to distribute or something else like possible gang affiliation.

I come from a police district where that that probably just means they're non-white.

Presumably this "new standard" GP mentions would be codified into law. And the law would still be 'paramount.'

I think procedures to effect expunging records follow the requirements of paper. If you have tens-/hundreds- of thousands of paper files, a law to require physical removal of the specific convictions is burdensome to whomever maintains the archive. A law that simply dictates ignoring old convictions under certain circumstances is bound to be misused (whether intentionally or unintentionally.)

Since we're now in a digital age of keeping these records, I believe that automating the 'deletion and cleanup' of files under specific conditions is laudable.

Addressing your comments about 'only charged ... but suspected of': this reeks of discrimination masquerading as safety. If you can prove it, charge the suspect and give them their day in court. If you can't prove it, a LEO's own suspicions have no place in the decision making process of other cases.

> Because at the end of the day, while we might not consider what they did a big deal now, when they committed the act it was a serious crime

"Serious crime" Tell that to people who were in possession of some cannabis flowers and are now sitting next to rapists and murderers in federal prison.

Hacker News is a gang affiliation. FANG employment is a gang affiliation.

You don’t want a criminal justice system where any decisions of consequence are made “behind the scenes automatically” no matter how solid your intentions.

I think it is a matter of practicality and old systems. Not to mention the headaches if something is supposed to be auto-done but wasn't.

I think the legal system could use through computerization personally but I am biased and recognize that would be one hell of a project to transition and avoid disaster - let alone the political capital involved.

It would be a bueraceatic task in a non-perojative sense - it is dull and not something impressive but if you are stuck doing it anyway it could make things easier for you.

>then the record of X conviction is considered expunged

What exactly do you mean by "considered," here? By whom/what?

The software mentioned in the article is Clear My Record [0].

[0] https://www.clearmyrecord.org

Ah, yet another example of a switch statement being "AI".

I think that's fake code in the images ;)

Well it’s relevant to the story, so if it’s fake they put a surprising amount of effort into it...

I was trying to work out what language it is - also switching on strings like that doesn't seem like a great idea.

The project is hosted on Github [0], I'm not sure if the entire source code is available though.

[0] https://github.com/codeforamerica/autoclearance

I would hope the entire source is available. We should always be wary of secret algorithmic decisions in court cases. The algorithm should be openly available for the public to see in the same way that the law is to be openly available.

Hi, engineer on the project here! That's an old version of the code that turned out to be too slow for the volume of records we need to process. The current version, which you see in the images from the article, is written in Golang.

You’re doing incredible work. Keep it up!

(Also use consts so you don’t make typos - or so your typos are at least consistent :)

Can you go into that a little more - the switch to Golang and the old version being too slow?

The code in the screenshot is valid Go (switching on strings, no parentheses around the if condition, incrementing a map value using ++) and could even be the output of go fmt (with the { on the same line as the if).

The app itself appears to be written in Ruby (the github link is posted elsewhere in the comments), and looking at the image included in the article further down I assume the code is from a Go app that takes the Ruby app's logs and generates some statistics.

Yes, that was painful to see. The strings are even capitalized in a particular way so it doesn't look like it had already been processed to normalize the formatting.

If you go to 1:51 in the video, you can see the same code on what appears to be an engineer's screen. It seems like it would be a lot of effort (and questionable journalism) to fake that.

I don't see anywhere in the article where they call it AI. It is described as an algorithm, which seems appropriate to me.

In the caption of the video "WATCH: AI helps clear cannabis conviction backlog".

The word is associated with the cutting edge whereas the actual code itself I imagine is mundane and hacky

The software does perform text mining (presumably via some sort of NLP?), an area that is firmly a subset of AI.

It could just be very simple keyword matching, especially if this is looking at court records.

It seems that this might indeed be the case. After writing the comment I’ve had a look through the (obsolete) Ruby code base of the project that somebody linked, and it doesn’t actually perform (what I’d call) text mining as far as I can see. Instead it parses the court documents, which seems to be a mostly structured format. No natural language processing.

for case in cases: if "cannabis" in case: case.flipverdict()

There u go ,crazy machine learning algorithm :O

Now you've created bugs where you overturn "not guilty" verdicts, expunge verdicts of cases that tangentially involved cannabis, and fail to expunge verdicts of cases where cannabis wasn't in the text but that still fell under the law change.

Haha if it's not correct to start with it must be some sort of "machine learning".

Besides, how many innocent verdicts for these crimes can there be? Prosecutors don't care near as much about pot possession as cops do (because they're not the ones getting the federal money), and they plea out possession when there's any doubt about convicting on more serious charges.

Minor point: the headline keeps it quite neutral as to whether this is a good thing or a mistake!

That's the point. You click on the article to find out.

That's probably a good thing overall... I mean neutral reporting and titling of articles. Too much reporting has become editorialized.

It is not a good thing in this case. The mere suggestion that it might be a mistake is editorializing, not being neutral to the article. I assume the comment above is being diplomatic in it's wording, but being "neutral" about whether these convictions were intentional or a mistake is not the same thing as neutral reporting.

The topic of AI or algorithms making mistakes is not relevant to this article at all. It doesn't summarize the content, and the implication that it might be code run amok, in the title, is super misleading and undermines the positive things the people in the article are trying to do, both the politicians who did the right thing, and the efforts of Code For America.

I wasn't just commenting in this case. I just mean in general, I'd prefer to see more neutral reporting and less editorializing.

Wasn't this a maguffin in one of the Batman movies?

my thoughts exactly lol

Is this really a sophisticated algorithm or simply a script or database update statement. Every piece of code as reported by the news these days is algorithmic or AI related

Imagine having a switch fallthrough bug in code like this…

No break statements!

Well done George Gascon!

Reply on news.ycombinator.com