I'm conflicted about this for a few different reasons:
(1) I am always concerned about start ups with hugely impactful products that seem to lack rigor around their technology. This one will be part of determining who gets what care or who gets sent to jail. Where's the evidence that Promise will make this decision better?
(2) It seems like streamlining and increasing the ability of governments to monitor parolees. Is a warrant going to be automatically issued to someone breaking curfew because they're visiting their sick mother?
(3) It lowers the barrier into the criminal justice system. Promise says they are designed to help the criminal which makes it more attractive for judges to place people, especially juveniles, into the system.
It seems like the founders want to make a positive impact on society, but there's a whole lot of slippery slope and other ethical issues here.
(1) We believe scale will be critical. Also, qualitative and quantitative analysis. I do not think this can be solved only through technology.
(2) Parolees are post conviction. We are focused on people who are incarcerated pre trial because they cannot afford bail.
(3) This is not what we said but really helpful to understand that is how it reads. We will need to think about it. Fundamentally, we are trying to make the system more just. People who can afford bail can get out, only people with less money usually stay incarcerated when they are bail eligible.
> (1) We believe scale will be critical. Also, qualitative and quantitative analysis. I do not think this can be solved only through technology.
This is the cavalier attitude that concerns me. If we are talking about predicting what wines people like than saying we believe scale will be critical is fine. When talking about something that will have a major impact on the lives of unwilling participants, you'd like to see at least some rigor.
> (2) Parolees are post conviction. We are focused on people who are incarcerated pre trial because they cannot afford bail.
People being Parolees or on bail doesn't make a difference to my point.
> (3) This is not what we said but really helpful to understand that is how it reads.
You don't think the care plan Promise creates will have a positive impact on those that go through the program?
What I found specifically cavalier was the statement that they "believe scale will be critical". That sounds like every ML/data driven start up that doesn't have enough data to actually make accurate models. There's a lot of fake it til you make it in startups like these.
What rigor would look like to me is something like:
(1) Do a historical analysis of who skips court and why.
(2) Come up with a plausible intervention strategy to solve problems identified in (1)
(3) Test your strategy in a well designed pilot and calculate it's effectiveness
First priority should absolutely be to get rid of the bail system, which is horrifyingly slanted towards imprisonment of the poor, and doesn't accomplish much of anything beyond that.
Your calendar thing might be useful. But not as a part of the bail system. Bail must go.
I think it makes sense to have multiple possible ways to compel someone to behave as you'd like. Money can be a way, but I agree that other ways could be as or more effective and accessible.
Bail or no bail, I do not see how your system will prevent the Shkreli and Holmes of this world to not be the prime benefactor and beneficiary of your system.
They would probably choose to get out without our program. The system works better for people with money. We are trying to bend the arc of justice but it is still bent.We are working with a population that is incarcerated because they cannot afford bail.
It looks to me like they're designed to help the accused, not the criminal. Here's a great infographic about the problem: http://www.pretrial.org/the-problem/
I think you may be right. They use the terms Jail and Custody instead of prison and incarcerated.
It's not that clear from the information presented though.
I also assume that they intend to be presented as an alternative to a prison sentence, if the judge is so persuaded. Just having a track record of compliance with Promise before any sentencing happens would probably be quite helpful.
I think this is one of the neater startups I've seen. However, privatized jails/prisons seem to be rife with abuse... privatizing parole seems like it would suffer the same temptations.
Very good feedback. Our site is focused on our clients who use very specific language. We will think about how to speak to folks who are not in the system right now.
> (1) I am always concerned about start ups with hugely impactful products that seem to lack rigor around their technology. This one will be part of determining who gets what care or who gets sent to jail. Where's the evidence that Promise will make this decision better?
I'm not with Promise, but let me tackle this one.
What currently decides which suspects stay in jail? Two things: a judge that sets (or denies) a bail price, and the ability of the suspect to pay that price, either directly or through a bond. What this means is that, as mentioned in the OP, the poor, who will be most strongly affected by being unable to work before their trial, are also those who will be most likely unable to pay their bond.
I would also like more information on how Promise will determine who will be eligible for their services, but whatever the process is, if it means that any number of people are able to continue working and helping their family as opposed to rotting away in a concrete box just because they can't pay to bail out, it's absolutely a net win.
Thank you. This is a beautiful answer. We agree on the criteria for evaluating what someone's risk is. Currently, it is primarily decided by the court. We are learning and adapting our own tool based on tools such as the one by the Arnold Foundation but also using qualitative analysis.
Some lit, Stanford CS paper: "Even accounting for these concerns, our results suggest potentially large welfare gains: one policy simulation shows crime reductions up to 24.7% with no change in jailing rates, or jailing rate reductions up to 41.9% with no increase in crime rates." Judges are famous for not even being close to the pareto optimal curve.
It's not a net win if, for example, someone is let out on bail because Promise calls them a low risk and they rape someone.
I agree that there is a huge need for reform in the system. My point is what evidence is Promise relying on when they promise to make improvements? Why should I think they're better and how are we going to verify in 1, 5, or 10 years that they are?
> It's not a net win if, for example, someone is let out on bail because Promise calls them a low risk and they rape someone.
So if a bail bond company grants a bond to someone and they rape someone, is the bail bond company at fault? Or the judge that permitted them to bail out?
And I'm willing to bet that for each person who commits a serious crime while out on this program, they will have hundreds, possibly thousands of people that don't - that are able to continue working and providing for their family when they otherwise wouldn't. Still an absolutely massive net win.
We agree. It is also really troublesome to make the case that poor people who cannot afford bail should be kept incarcerated for public safety, but people who have the money to bail out, can await trial at home.
The people that make bail have a financial incentive to “be good”, and not forfeit their bail. That part of the system works fine.... if they offend while out on bail, Uncle Sam gets paid. It’s a broken fix for a broken system.
Harsh(er) punishments are an effective deterrent, and if they don’t deter they should be harsh enough to wipe out offenders through attrition. Eg, shoplifting with 2 hands is easy, with 1 hand is difficult, and with 0 hands is nearly impossible.
Yes, absolutely. It's part of the judge's job to make some initial assessment of risk and set bail accordingly (or deny it entirely). For example, if the charge is murder, bail is automatically denied.
> if the charge is murder, bail is automatically denied.
Not at all. Some judges may have that policy, and perhaps some state laws require it (though that would arguably not be constitutional and a violation of the separation of powers), but it's not certain that someone up on murder charges would not be able to bail out.
At any rate, the point is that if someone who has been charged - just charged, not convicted; remember, we're all supposed to be pretending people are innocent until proven guilty here - with a crime commits a crime before their conviction, the fault lies only on that person and that person alone, not on anyone else - and certainly not the judge.
That example was my understanding, glad to be corrected.
> At any rate, the point is that if someone who has been charged - just charged, not convicted
Agree completely. But if we are not making some estimation of the possibility of guilt and possible behavior by a guilty party given freedom then what is the point of ever denying bail? We deny bail because they might be guilty, and if they are it might not be safe for them to be released. We're not saying that they are guilty, but conceding that it might be negligent to allow them freedom before determining their guilt.
> the fault lies only on that person and that person alone
I'm certainly not trying to absolve the perpetrator, but certainly from an ethical point of view at least (IANAL) you have to admit the possibility of negligence.
This is not the point of bail. Bail is not some sort of speculative conviction that you can pay to avoid. The only purpose of bail is to hold somebody if they may be reasonably judged to be a flight risk.
Bail is, btw, almost certainly unconstitutional and should be abolished. It's one of those things that slipped through the cracks and was kind of grandfathered in. The Supreme Court has ruled again and again that in a system where persons are presumed innocent any sort of pre-conviction punishment by the state (fines, excessive jail, forced hospitalization) is not allowed. Somehow though the judge has the power to declare somebody a flight risk and order them held until trial. This is a little mitigated by the right to a speedy trial but is still likely wrong.
Yes, the Constitution implicitly permits bail by prohibiting excessive bail. (That's at least one way to read it.) The courts have also asserted that bail is somehow "fundamental" to the system of law (it's not to be questioned). Still what's not clear is (1) whether the power to deny bail is constitutional (some might think refusal of bail, in effect a bail that cannot payed at all for any amount of money is rather excessive) and (2) whether excessive should be understood as relative of the client's ability to pay. In fact Stack makes it somewhat clear that the defendant's ability to pay partially determines excessive... and yet here we are: every year millions of people go to jail because they are denied bail or they cannot pay. It should be clear that the current system where the government locks up millions of defendants because they cannot pay bail (or are refused jail) is not what was intended by the framers. Abolishing bail or ensuring that bail is always and everywhere affordable and reasonable is the way to go. A startup like Promise offers a perhaps much-needed band-aid but the entire system is broken and should be revisited.
> Promise offers a perhaps much-needed band-aid but the entire system is broken and should be revisited.
That's actually my biggest fear with private services like Promise — a new for-profit bandaid both reduces the pressure to reform the underlying system just as it is gathering real steam, and creates a new set of parties with a profit interest in preserving the underlying system.
> Agree completely. But if we are not making some estimation of the possibility of guilt and possible behavior by a guilty party given freedom then what is the point of ever denying bail?
Bail, or it's denial, is principally about the risk of the accused not showing up to court, and thereby escaping legal process; the type of crime is also considered, but again that's mostly (but not entirely) because it factors into risk to the process (both motivation to avoid process and risk of violence directed at witnesses, etc.)
> So if a bail bond company grants a bond to someone and they rape someone, is the bail bond company at fault? Or the judge that permitted them to bail out?
It's not a matter of fault. It's a matter of who is going to do a better job at granting bail. Promise or a Judge?
Promise says that they can, which in most tech start ups is enough. If we find out 5 years later that your algorithm sucks at, say, picking out clothes for someone, who cares? You go out of business and some people will have gotten worse clothes. If Promise's algorithm sucks then in 5 years later we will have a bunch of additional crime victims on our hands.
>And I'm willing to bet that for each person who commits a serious crime while out on this program, they will have hundreds, possibly thousands of people that don't - that are able to continue working and providing for their family when they otherwise wouldn't. Still an absolutely massive net win.
How will we know you are right? The things that Promise does are the things that should end up as peer reviewed journal articles. That's the point of my question. How is anyone going to know that Promise is working?
> It's not a matter of fault. It's a matter of who is going to do a better job at granting bail. Promise or a Judge?
Promise would not be replacing the function of a judge here. That's ludicrous. The judge would be permitting a suspect to work with Promise's system rather than rotting in jail.
> How is anyone going to know that Promise is working?
When people are able to work and take care of their families who would otherwise be sitting in a concrete box, waiting for time to pass.
> Promise would not be replacing the function of a judge here. That's ludicrous. The judge would be permitting a suspect to work with Promise's system rather than rotting in jail.
Even if their risk assessment plays no role in who gets bail, they still are involved with major decisions impacting people's lives so the point still stands. Which, after three posts you still haven't really addressed. What evidence does Promise have that their product will make things better? This isn't really something you give the old college try and hope to find success.
>When people are able to work and take care of their families who would otherwise be sitting in a concrete box, waiting for time to pass.
And how are we going to know that is happening? This is the type of stuff that is done with controlled experiments resulting in peer reviewed journal articles.
I'm not sure if you're unclear of the system, or have some specific point I've been unable to ascertain through your prior posts, but as I understand it Promise is not an alternative that allows more people to be released that otherwise would have been detained, it's an alternative that allows people that the judge already deemed eligible for bail to have a non-monetary option for release.
That is, if you have resources you can get out without Promise, Promise just helps those that were already deemed not a risk to the community by the Judge find a method of spending the weeks/months leading up to their trial still being able to lead a semi-normal life and not incarcerated and not have to have significant financial resources to do so.
You're right that it is somewhat danced around in the initial announcement. I think a lot of people, including myself, read into it based on some hints throughout (such as where it talks about pre-trial incarceration along with some other types), and made assumptions.
That said, there is a comment that clarifies that this is at least one area they are looking at:
The fundamental issue is that we are focusing on poor people who are bail eligible and incarcerated only because they cannot afford bail. [1]
That said, they could try to expand beyond that. In addition, this additional comment means a lot of what I said, and others have said, may not quite be accurate, at least in the future:
It is a judge. However, there is a recent court decision(Humphrey), that is creating a lot of change within the system. [2]
You are correct. If someone can afford bail, they will likely bail out. If someone is not given bail by a judge, they will not be allowed to bail out or participate with Promise-they will stay in jail until the resolution of their case. If someone has been given a bail, but cannot afford to bail out, this is someone who is eligible to work with Promise. If the court agrees, they will be released from jail and returned to the community (their job, home, family, etc.) without posting bail. Promise will then help support them to 1. Comply with their court mandated obligations and 2. Connect them to services based on a needs assessment conducted by Promise staff. I hope this is clear. We will work to review the website and make sure that it is. Thank you!
> Which, after three posts you still haven't really addressed. What evidence does Promise have that their product will make things better? This isn't really something you give the old college try and hope to find success.
I've addressed it. People who cannot afford to pay bail (or for a bail bond) will be able to work and be with their families rather than stay in jail. Do you disagree that this is "making things better?"
>>When people are able to work and take care of their families who would otherwise be sitting in a concrete box, waiting for time to pass.
>And how are we going to know that is happening?
Are you over-thinking this? We will know this is happening when people show up for work and then go home to their families at the end of the day.
>I've addressed it. People who cannot afford to pay bail (or for a bail bond) will be able to work and be with their families rather than stay in jail. Do you disagree that this is "making things better?"
It's not evidence. You're answer defines a success criteria, but does not give evidence. I agree that if fewer people stay in jail that would be a success. What proof does Promise have that they will achieve that?
>Are you over-thinking this? We will know this is happening when people show up for work and then go home to their families at the end of the day.
And how am I supposed to know that is happening? Is Promise going to publish numbers? What is their control group?
But to the extent that any one of those metrics may show that your methodology has a negative or even negligible impact, what mechanisms exist to ensure your fidelity to these metrics, instead of leaning on public relations and marketing that pick up the slack?
No one is here accusing you of doing any of these things. On the contrary, there is an implication that you are positive that you will be able to do these things, even though you're obviously still validating the idea, at some level.
The fundamental issue is that we are focusing on poor people who are bail eligible and incarcerated only because they cannot afford bail. Yes, some people who have been accused of a crime, do commit an additional crime. However, it is fundamentally unjust to only incarcerate those who are poor and mostly black and brown.
Someone who hasn't been convicted of a crime is not yet a criminal, at least not from the perspective of a criminal justice system based on "innocent until proven guilty." If we start restricting the rights of people we're merely suspicious of in the off chance they commit a crime, we create a very dystopian society.
I'm pretty sure the government would be the final arbiter of who would be let out. Which does raise the perennial issue of, no politician wants to be the person to let that one guy out who commits another crime, even if the other 9,999 did great.
Consider the Lucas Critique [1]. To paraphrase, the general idea is that it is natural for organizations to seek to optimize their ability to navigate any regulation, or any impediment to maximizing their profits, or otherwise maximally seeking their ultimate goal. Its advice is that it is counterproductive to try to reign in the behavior of sophisticated organizations with exact targets of compliance, as you are simply giving those very organizations a blueprint for the necessary optimizations for both minimal, often semantic compliance, and a continuation of the actual behavior the original regulation intended to curb.
The application to this subject is the very idea that companies such as this are motivated to optimize their methodology of convincing judges or whatever ultimate adjudicator exists, to exonerate their clients.
I don't see any indication of Promise being a nonprofit, so we have to assume their primary goal is in fact profit.
Bad actors shouldn't be able to throw money at the optimization of skirting the law. To the extent that Promise is successful, that conclusion can and will be easily drawn to great effect.
Will every employee of Promise need to convince their state bar association of their character and fitness?
Promise isnt (yet) working with convicted clients. This is purely a pre-trial intervention, as they keep saying. So supposing that they will be trying to exonerate their clients is a red herring.
Supposing that their motives are pure profit is extra cynical; there are plenty of organizations in the world that do good and still earn a little money.
Since this is all pre-trial, the incentives for Promise are to ensure that they have the best rates possible of a) showing up for trial and b) not committing extra offenses until then. The only danger I see is the possibility that they would over-select candidates and get too many who use this as a way to dodge their trials or commit more crimes. Since either of these harm their ability to do business, I would think you agree that they are incented to minimize those events.
I dont think Lucas applies; its about the weakness of regulations to control actors. This is a new organization that aims to offer an alternative for existing government actors.
This is all happy-path talk. We'll do everything right forever in all cases no matter what, talk.
I'm concerned about what their plan is for if and when they find out that the hypothesis that is their most prominent, and most marketable use case has been falsified, yet it turns out that some other market they hadn't even considered at the start is absolutely printing money for them, even if it represents significant mission creep from keeping poor people out of jail.
Supposing that their motives are profit driven is putting two and two together. We're currently in the middle of YC demo days. This very post constitutes a prong in their launch strategy.
Claiming that profit motive and the motive to do good in the world are mutually exclusive is the true red herring here. It is separately egregious, because it also aims to frame my argument as wanting them to fail in any related venture. As I've said all along, I'd simply like to know what, besides blind trust, should we lean on, to ensure that a private company, who will undoubtedly have lots of reasons to want to keep its methodologies a secret, will be honest and forthcoming in the better than average chance that their favored marketing use case doesn't end up as what is truly sustaining their business. After all, no plan survives first contact with the enemy.
The answer to people claiming you might be a bad actor is go the extra mile, and do things and con men cannot, such as provide meaningful context, and coherent arguments that actually address your skeptics' questions.
The Lucas Critique is actually one of those theories that has been put forth in many permutations, such as the Cobra Effect, Campbells Law, and Goodhart's Law. In plain english, its just saying that if you want to prevent some behavior in others, giving clear, explicit, comprehensive targets of what it means to comply is counterproductive, because those targets are always an imperfect proxy for the actual behavior you wish to curb, and "compliance" can nearly always be achieved to the letter of the law, while not in the spirit of the law. It applies to social theory in general, which is the context of Campbell's Law.
I think that your heart is in the right place, but objecting to something like Promise just because they have a profit motive is letting perfect be the enemy of good. As I said above, if this system means a single person is able to be working and taking care of their family rather than sitting in a cold concrete box, a great justice has been done to the world, regardless of the motivations of those responsible.
I think your optimism has a fighting chance of being mirrored in reality at some point in the future. I don't object to the idea of Promise. I object to absolutist statements claiming that companies who plan to do great good by carrying out operations that could easily be subverted to great harm are unequivocally a net positive.
If what you're saying actually comes to pass, I'll celebrate it. But to gas light the devil's advocates who simply want to ensure that this isn't another group of Elizabeth Holmes style charlatans, is transparent and fully deserving of being called out as grade A, corn fed bullshit. I can almost hear the raspiness in your voice as you describe these situations. There's a Sean Colvin song playing in the background.
One person's terrorist is another person's freedom fighter. How will you deftly strum my heart strings if its an objectively guilty rapist who is allowed to take care of their family, rather than sit in the cold concrete box where other possible victims are safe from him?
Your questions remind me of how Eli Whitney's cotton gin, originally invented for the purpose of making the lives of slaves better, instead dramatically increased the demand for slaves causing the enslavement of millions.
With regards to point #1, based on reading the description of Promise and glancing at their website, it does not sound like technology plays a meaningful role in deciding who is released from jail.
> Here's how Promise works: We work in partnership with governments who release people from jail on condition that they work with Promise as an alternative to being in custody. We also provide support to people under community supervision. We use an intake assessment to create an individualized plan that is based on the risks and needs of each participant."
I took that to mean that someone from the government and someone from Promise look together at the record of someone in jail (and obviously the government would have the final say), decide if they are a candidate for release.
And, my first thought about the "intake asseessment" step to ascertain the level of risk of individuals in the Promise program was that it would be like any other social science/psychology-based interview and questionnaire process.
#2 is a good point, and a concern I had as well. At the same time, if some progress could be made (as Promise intends) on decreasing our prison population, that would be really good for society.
#3 seems like an assumption. Maybe, maybe not - it's hard to say without input from those who have experience w/the criminal justice system.
I definitely agree there are a lot of ethics issues here. That said, on a scale of 1 - 10, for tech companies/ideas with potential ethics issues, this is way below the Google/Facebook/Palantir/etc level - i.e., this is not the slipperiest slope we have to slide.
When you are in jail or someone you love is in jail the choice to get out with a smartphone app and support is pretty clear. It is interesting feedback about monitoring but when it is you or your family, you want out. The most folks reaching out to us, have families who are trying to help their loved ones.
While I agree that limited release is better than incarceration in many instances, I don't think the GP should be dismissed so quickly. The third point is important. If being put into this system is viewed as not has harsh as being incarcerated, does that change how likely people are to have charges pressed? Does this bypass most of that by being associated with the pre-trial but post-charges phase? If it ends up causing a 20% dorp in pre-trial incarceration but somehow encourages 20% more arrests that have charges applied, is that worthwhile?
I don't know the answer to any of these, but I think it's worth discussing. As such, I've decided to view your dismissal as a compliment. "Only on HN is app-based remote monitoring as a method to reduce jail seen as a complex system with many inputs and outputs that sometimes have unintended consequences and as such it deserves in-depth discussion, especially since it's such an important issue." (not that I think this discussion is only available on HN)
> Only on HN is app-based remote monitoring seen as potentially worse than being IN JAIL.
That's a bit of a strawman - there are other ethical issues at play here. Some people don't have smartphones. Others will have theirs lost/stolen/broken. Do they go back to jail?
Is it a local thing to have incredibly cheap smart devices? I can go a town over in the rural US and get a cheap smartphone for like $30 with a prepaid plan. Granted, I can't imagine it would be an amazing experience to use, but far better than jail.
$30 is a lot for some people, especially those facing jail. Then there's paying for data. The app might not work on the older Android version that's on a $30 device. People with intellectual disabilities may struggle with a mobile phone.
None of these things should wind up determining whether a person goes to jail or not, but that's a risk here.
Creating another layer of bureaucracy, even when covered in the veneer of startup tech, can definitely make things worse. Here are some open questions that may show how that's possible:
* What will happen if nefarious actors discover how to hack or game this system? Will you be able to pay the mob to get hacked out of jail?
* On the flip-side, could a nefarious actor put you in the penal system, or at least make it look like you were arrested?
* What protections are there around malfeasance within the company? Parole officers are public employees, but what's the liability within a LLC?
(1) I am always concerned about start ups with hugely impactful products that seem to lack rigor around their technology. This one will be part of determining who gets what care or who gets sent to jail. Where's the evidence that Promise will make this decision better?
(2) It seems like streamlining and increasing the ability of governments to monitor parolees. Is a warrant going to be automatically issued to someone breaking curfew because they're visiting their sick mother?
(3) It lowers the barrier into the criminal justice system. Promise says they are designed to help the criminal which makes it more attractive for judges to place people, especially juveniles, into the system.
It seems like the founders want to make a positive impact on society, but there's a whole lot of slippery slope and other ethical issues here.