Whistleblowers: Inflexible prison software says inmates due for release should be kept locked up behind bars Prison inmates in Arizona who should be eligible for release remain incarcerated because the state's inmate management software can't handle sentence adjustments, it is claimed. According to public radio station KJZZ, unidentified whistleblowers within the Arizona Department of Corrections revealed the software problem, which …
If that report is anywhere near accurate, as well as the ridiculously high cost per hour of bug fixing, 2,000 hours to tweak a calculation?
They evidently view Arizona's State Government as a bottomless pit of money who'll willingly throw millions of dollars at them - possibly even billions if they're charging over a thousand dollars per hour and over a thousand hours of development time to fix each of tens of thousands of "unresolved issues" and "feature requests"...
> If that report is anywhere near accurate, as well as the ridiculously high cost per hour of bug fixing, 2,000 hours to tweak a calculation?
Yeah, just a tweak to the code. And some testing to make sure you don't release a murderer by mistake - could you knock that off in half an hour or so as well?
> with proper unit testing, yes...
Repeat after me: unit testing is not system testing.
Please explain how a unit test will catch the scenario where a prisoner has transferred prison 5 times; been previously released but re-incarcerated after committing a further offence; has undertaken 3000 hours of study; but 10 days of which was on a non-contributing course? And then catch all the other nCr permutations - because a judge *will* ask you to explain if a prisoner is released early by mistake. And your explanation better be better than "well, we ran a unit test".
Because this sounds like a new requirement brought in by recent legislation, it's probably more like implementing a new metric that does not currently exist in the system along with all the associated changes to screens, queries and updates, and then incorporating that metric into the release date calculation.
How much does adding an extra field to a database cost? It depends on the database system, but I thought from my relational database training forty years ago that this was one of the features of a RDB that I'm sure cannot have gotten lost in the time since then.
But testing would need to be done, and the queries and update methods would have to be written, along with the required auditing, so it could well be quite expensive.
Which came first the software going live or the law SB1310 being passed as they both happened in 2019?
Even if it went live earlier (not earlier then six months) than SB1310, they must have know it was on the books to be debated. Should they not have delayed the software going live until it could handle changes to the release credit formula?
You'd think so.
The bill had its first reading in Arizona's Senate in January 2019 and it sailed through the process: a 60-0 final vote in the House, 28-2 in the Senate, signed into law by the Governor in June 2019.
ACIS (Arizona Correctional Information System) went live in November 2019.
I see this all the time.
The software would have been written to meet a spec and rolled out many months after approval. Once the scope, design documents, implementation and user acceptance testing has all had final sign off, it’s way too late to change things.
To adjust the system to handle the new rules would be chargeable out-of-scope works which the company could put any price on, knowing that hiring a third party to do the changes might cost less initially, but more in the long run to take over support.
Additionally, it’s in the best interests of Arizona State (not the programmers) to keep quiet about bugs to avoid being sued into the ground for mistakes made as a result of keeping people locked up for longer than they’re meant to. The development house will have a couple of executives laughing their arse off at this situation, enjoying all the easy money...
I agree it would be too late to change things and get the new formula supported but going live could have been delayed until it had been fixed.
The blame is on Arizona State letting it go live when they knew that it could not handle then new formula. Yes it would have cost them and been late but it could cost a lot more if it ends up in law suits.
It seems the requirements were poorly speced not allowing things like the sentence reduction formula to be adjustable. But hey hindsight is a wonderful thing. :)
I disagree it should be doing that job.
"The software, ACIS (Arizona Correctional Information System), implemented in 2019 at a cost of $24m by IT biz Business & Decision, North America, is said to contain a module for calculating the release dates of inmates.
The module's code, however, hasn't been able to adapt to Arizona Senate Bill 1310, a state law signed in June 2019 to allow non-violent inmates in Arizona to earn credits toward early release as a reward for participating in state-run education and rehabilitation programs."
So yes I read the story and stand by my statement "It seems the requirements were poorly speced not allowing things like the sentence reduction formula to be adjustable. But hey hindsight is a wonderful thing. :)"
Because how can you calculate release dates of inmates if you do no allow for the early release program which existed before Bill 1310 and also allow the formula for earning credits to be adjustable.
"The development house will have a couple of executives laughing their arse off"
... and burnishing the bottom line they report to the head office execs who report to their owners who report to theirs and so on until the execs at the top know nothing about any of it. (Business & Decision, North America is the US subsidiary of a French company owned since 2018 by a subsidiary of an arm of Orange.)
Mm! So should I hardcode the tax rates in this payroll software, and ensure an income for life, or give the user an admin option, so they can define and redefine to their heart's content?
Damn! I'm just too professional about my work to let grubby thievery get the better of me!
As a note, I always code with the knowledge that things change, and always create admin screens that give the user the option to make those changes themselves, and I always supply full documentation, often displayable on the screen to assist the admins in implementing those changes.
Yes it is hard work trying to anticipate how "constants" might change, and yes it's a pain in the butt trying to create as simple an environment for the user to navigate, but anything less doesn't ship with my name on it.
I was once offered a half-time job producing a series of similar bespoke programs. I expressed my surprise, explained the principle of code/data separation, talked through the real requirements, and spent three months producing a single configurable program which put my client's previous vendors out of six months' billing per annum for (if they were half-way competent and had done essentially the same thing behind the client's back) about three days' work.
Well, yes... but how long does it take to (a) get a hearing and (b) prove to the judge's satisfaction that the discount to the sentence is earned? You _know_ the Arizona Dept of Corrections will produce evidence that their "computer says no", so now the convict needs to prove that the DoC is wrong...
They do state in the update that there is a manual system that ensures no one is imprisoned when they should be released, and that it has been litigated in Arizona may times and always supported by the courts. So maybe someone over there had enough common sense to ensure a manual work around in case of IT failure?
Very dodgy ground using an IT system to dictate release dates anyway as it's not a straightforward process.
In England and Wales whilst the national database (Nomis) does do calcs, these have to be manually checked at various points in the offenders sentence. Prior to that the previous system didn't do date calculation at all.
Don't know about that but every project I've worked on in the last few years has had a contact with a clause in it requiring the supplier to incorporate any legislative changes, either as part of the routine software update releases or as a an out of cycle release. No change request required.
Normally when doing the work for free when legislation changes, both the supplier and the customer have no control over the legislative changes.
In this specific case, the customer gets to decide the legislative changes. That would be unfair on the supplier - though it looks like the rates they charge could provide a bit of a cushion for this.
There are two proven ways to reduce recidivism:
1) Never release prisoners.
2) Use the incarceration to rehabilitate prisoners, support them on release.
Too many people think that 'criminal' is a sub-species of humanity and that they are beyond hope. These are the 'lock 'em up and throw away the key' brigade. Despite centuries of evidence they persist in thinking that locking someone up will somehow stop them doing it again on release.
Anyone who understands human behaviour knows that locking someone up makes them more bitter and resentful. What's needed is a serious attempt to treat 'criminality' as an illness. Most criminals can be changed if enough effort is put in.
Never realising prisoners only dictates on whom they can practise their criminality. In fact there is a crime in the UK of 'Prison Riot' which you can ONLY commit if you are in prison.
Supporting them on release is a good idea, but that depends a great deal on the society into which they are released. Would you go for the divided, racist, class-ridden prejudice of parts of the USA, or the enlightened supportive, and far less prejudiced Norway? (Norway has an admirable record of rehabilitation, but spends loads of time effort and money on treating prisoners as human beings, see Rutger Bregman's "Humankind", ISBN 978-1-4088-9893-2).
> Most criminals can be changed if enough effort is put in.
Aye, and there's the rub. Most governments would rather spend more money on locking them up ('being seen to be doing something') rather than spending a lesser sum trying to prevent them from re-offending. Not good for the votes, you see.
I have just listened to the BBC Radio 4's 'The Life Scientific' today. It was about psychosis. Apparently someone who experiences a traumatic event before the age of 16 is 3 times more likely to have a psychotic episode than someone who has a happy childhood. Also the worse the trauma the more likely the episode. Spending money on treating people with a troubled childhood pays dividends.
https://www.bbc.co.uk/programmes/m000sj7c
A previous episode about mentalising was also fascinating:
https://www.bbc.co.uk/programmes/m000dpj2
Note: user account required to listen. Also some bad events are described.
Check out Eclectic Man's recommendation - Humankind.
The analysis that has been done, and IIRC is being done by some red states in the US after looking at the Scandinavian model, is that in the long-term it is cheaper to put the money into good prisons and into proper rehabilitation. With recidivism rates of only 20% in Norway against 50-70% (depending on source) in the US, it just makes sense in the long run for all except the private prison owners...
A common theme in Norwegian prisons is that they are looking after people who will be released and will then be someone's neighbour, employee, etc. Better for all to integrate them back into the community properly!
Politically, it is easy to sell taking revenge on wrong-doers. This is something that people can see is happening, so we are seen to be doing something about crime. Criminals being rehabilitated, so they do not commit crimes any more, rarely makes the news. Yes, I know there are news stories about ex gang members working to prevent kids joining gangs, or whatever, but I suspect those people are far outnumbered by former criminals who sort their lives out, keep their heads down, and nobody knows about them.
This is the same state that wants to pass a law that gives the Governor/State legislature the power to overrule the will of the people in all future elections.
As the US Constitution preamble starts with
"We the People..."
They want it to be replaced with "We the GOP..."
because the state is solidly red but voted blue in 2020. The good ole boys can't let that happen again now can they?
"ADCRR updates the calculation multiple times daily to ensure appropriate release times are calculated and acted upon"
Given that sentences are only counted in days, how exactly does recalculating them multiple times per day help? If a person is not elligible for release at 0900, they're still not going to be elligible a couple of hours later.
I also note that they state the numbers of prisoners in various stages of participating in this program or being encouraged to do so, but don't appear to say anything regarding people who have been released earlier as a result. Overall, the whole response seems to use a lot of words to say absolutely nothing relevant to the actual question at hand.
"Given that sentences are only counted in days, how exactly does recalculating them multiple times per day help?"
It doesn't. All it means is that either (a) they run batch processes a couple of times a day (that counts as multiple) to process any changes that might have come in since the last batch or (b) they they process changes as they come in and there are normally two or more per day. Either way they're recalculating several times a day even if they're not recalculating them all. Presentation is everything.
Either that or they get a different answer every time they run the program.
They could also save a lot of manual (or software) processing by only recalculating days left every 2 weeks until someone is down to a week left or less (with new calculations - could change if the formula changes again) given that it is 3 days credited per 7 days served (so max 6 day change in 2 weeks).
I briefly did prisoner support once.
I gave a lift to a girl who was visiting her ma, as she did every week even though he was 60 miles away and travel was expensive. He was inside for four years for assaulting her. He'd only been sentenced to two years but he had to do an anger management course before his release. There was no anger management course at that prison so we paid an extra £88,000 to keep him inside. Maybe more than that, I never followed up on his case. She hadn't even wanted him prosecuted and was impoverished supporting him.
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
