Algorithms Designed to Fight Poverty Can Actually Make It Worse
scientificamerican.com
scientificamerican.com
http://donellameadows.org/archives/leverage-points-places-to...
Aka, "social mobility"
$1.16 billion is a lot of money to give to companies when it could pay for a good number of other things.
Edit: I don't literally believe this. I'm just saying that most people who have made it are not interested in everyone else making it to their level (that would be impossible, economics aren't zero sum but we can't all be Bill Gates). They're interested in making everyone else content.
Just about nobody gets far in capitalism subscribing to the the sort of zero-sum theory you allude to[1]. And even if: it'd be idiotic for them to base their decisions on such a theory, considering the "ruling class" tends to be rather old, while social mobility only has effects over several generations.
If you require empirical proof, look no further than the super-rich' historically preferred mode of philanthropy: the scholarship.
Scholarships are pure social mobility: take promising young people with limited means and allow them to attend elite schools. Surely the Rhodes Scholarship, Gates Scholarship, Schwarzman Scholarship, New York Times Scholarship, or any of the thousands of these programs would not exist in the cynical world you describe.
[0]: https://en.wikipedia.org/wiki/Schwarzman_Scholars [1]: I had a quip about current presidents being the exception here. But as it turns out, the Miss America Scholarship is among the largest in the US.
That conspiracy theory is immaterial to the discussion. The point is that social mobility and poverty are orthogonal concepts. You can have zero poverty but no social mobility, or high poverty and high social mobility. To put it differently, the reason why being born in a middle class family is a strong indicator that an individual will live a middle class life, with a low probability of becoming either upper class or lowe class, is not poverty.
> I'm just saying that most people who have made it are not interested in everyone else making it to their level (that would be impossible, economics aren't zero sum but we can't all be Bill Gates). They're interested in making everyone else content.
The assertion about "not being interested" is meaningless. I'm not interested in anyone winning the lottery, but that does not mean I don't want anyone to win or even that I intend to take any sort of action to stop anyone from ever having a chance of winning a prize. It just mean that I don't spend any time thinking about others winning lotteries because I have far more interesting and engaging things to do.
That's what we all do.
How much time do we collectively spend doing anything to actually tackle a hard problem such as hunger and war? Is that an indicator we want everyone to die from a combination of starvation and violence? Of course not. So, why is this vague notion of caring about somethong being conflated with being actively engaged in causing a problem?
It doesn't have to be completely free, but set in a way that motivates people take care of their health, requires them participate a bit to not overwhelm the system with their petty/untreatable issues, but not ruin them.
Just say free. Say subsidized. Say something less meaningless than accessible. Technically in the USA healthcare is 'accessible' right now. Problem is after I access it I can't pay for it.
As someone who's family has benefited (and pays for via taxes) from a well funded national health service I believe it represents significant progress in a nation's development.
How would you get a mortgage on a house, or save for your kids' education if any wealth you accrue risks getting re-allocated?
(Guess it depends on your value of X)
I think you're conflating social mobility with income growth. all social mobility means is how easy it is to change your social status in terms of wealth. while it's true that randomly redistributing wealth will disincentivise hard work, it doesn't matter because the shuffling will still change everybody's social status.
Otherwise the work, money, and time are going to be wasted.
As for the second part of your comment - what if you didn't need to go into debt, and education was free?
I think that's a legitimately interesting question.
Maybe the actors in that situation would attempt to build systems that don't rely on having outcomes based on individuals accruing wealth for themselves, but instead on collective assets that everyone has access to at all times.
Pragmatically, yeah, there are all kinds of problems with this concept, but for me your question does more to make radical, frequent, and random wealth reallocation interesting than it does to point out a flaw with the idea.
It's been ~10 years since I looked at the numbers, but I remember it being something like 30% of bankruptcy filings are due to medical issues and associated costs.
Maybe systemetizing it a priori would have a different impact on people's perception about the mechanism though?
According to this https://www.yourmoney.com/credit-cards-loans/living-beyond-m... circa 10% of bankruptcy proceedings were due to ill health.
I don't think much can be drawn from that though as there are too many legal and cultural differences in the application of bankrupsy.
[1]https://prescriptions.blogs.nytimes.com/2009/09/07/insured-b...
Also, ceteris paribus, if you get ill and you've worked harder, you'll have more resources to help yourself, so you'll be less likely to be broke by the end of it. Under this system, how hard you worked doesn't increase your expected amount of money next year.
Also, all the rich would immediately flee the country.
I was making a point about social engineering. Similarly: if you optimize for low inequality, you get Cuba.
Looking at someone's entire life and the likelihood that they will suffer poverty, their working wage is only one of many factors.
You need well paying jobs, low unemployment, and a social safety net for economic security.
Eubanks presents the impact that rigid, opaque systems and processes can have on individuals, and does so by focusing on troubling policy implications, but also by highlighting the suffering experienced by named individuals whose unique personal circumstances are examined. This is not only an effective writing device, but also demonstrates the empathy gap of software, and how an impersonal process -- whether intentionally or not -- serves as a tool for bureaucrats and low-level decision-makers to offload difficult negative decisions to an amorphous entity like 'the process'. I feel that this is a key, if not always overt motivation of most bureaucracies, and has universal applicability.
There's also the notion that when ML is fed with raw data, it picks up correlations present in the data regardless of whether their presence is due to a lurking variable. Later, making use of such a model for decision-making perpetuates the bias, and a haphazardly chained sequence of such processes increases the likelihood that a cohort's observed outcomes will self-fulfill in the future. Conversation on this matter in recent years has tended towards the social justice implications more so than the lack of intellectual rigor that allows this to be possible -- the former is a compelling point, but it's unfortunate that there's not enough coverage of about the harm of giving complex tools to those who don't understand them.
[1] https://virginia-eubanks.com/ [2] https://www.citylab.com/equity/2018/02/the-rise-of-digital-p... [3] https://virginia-eubanks.com/media/
I'm sometimes skeptical of human-narrative analyses, because they make it easy to launch equally-vivid attacks on a plan that fails 10 people and a plan that fails 10,000. But with rigor behind them, there's a lot to be said for them as existence proofs - especially when we're talking about the margins of a plan instead of its core. Knowing what the edge cases look like matters, and for that it's hard to beat talking to individuals with problems that don't fit on a form.
(As an aside, I think software often creates not only an empathy gap but a mechanical one. An analog system can often be forced to accommodate oddities like self-employment and unconventional address formats, but "the computer won't let me enter that address" is a brick wall. I've never found an investigation of this, but I suspect it's a major drawback of digitization.)
Your final point about ML is a vital one, and one I'm endlessly frustrated by. We're increasingly aware of issues like racial bias in parole-predictions systems, but there's alarmingly little focus on the fact that a system which does that is flawed in general, and might be making the same assumptions about e.g. low-income neighborhoods without anyone noticing. Unlike human bias, a flawed algorithm is equally happy to abuse any strange correlation it can find.
> Near the end of 2006 Mitch Daniels, then governor of Indiana, announced a plan to give the state's “neediest people a better chance to escape welfare for the world of work and dignity.” He signed a $1.16-billion contract with a consortium of companies, including IBM, that would automate and privatize eligibility processes for Indiana's welfare programs.
- Means testing frequently makes things worse, regardless of "algorithm"
- If you want to relieve poverty, don't give huge amounts of money to multinational mediocre consulting firms
I think this is optimistic even. At some point one evolves into a society where what determines wealth is one's aptitude at obtaining social benefit. On the high end, say, a CEO of a defense contractor like Lockheed Martin or Airbus (or more disgustingly American Shipbuilding, which billed for ships it never built) and on the far end, say a veteran suffering from paranoia and PTSD that refuses to obtain government help for, understandable reasons.
I only contribute to Amnesty International, but that's my personal choice (and they are in Geneva too...)
What works somewhat better is to condition cash payments on some simple and verifiable prosocial behavior, such as keeping your kids in school with passing grade.
Another (related) example would be access to mental health resources.
Waiting for someone to hit rock bottom before helping is obviously suboptimal.
And the premise here is that this was unintended?
I think I found the problem: credulous writers.
because it's explicitly allowed by the rules. I like your flair idea though, seems like a nice middle ground.
I concur. It's hard to have a discussion about an article that only a few people can read. The flair idea is a good one. We already require post titles on old articles to contain the year, seems natural to require disclaimers of paywalls.