> the type of browser and hardware used to access the internet, the history of online searches and purchases ... once powered by artificial intelligence and machine learning, these alternative data sources are often superior than traditional credit assessment methods, and can advance financial inclusion, by, for example, enabling more credit to informal workers and households and firms in rural areas
Here is one of the most sensitive applications of ML. This one, like face recognition, can directly harm people.
These guys only seem to think about the "superiority" of their method in minimizing lending risk. But the other party into this deal can also have something to lose.
What's the difference between this and the Chinese social scoring system? They want to use private search logs. "Your privacy for a credit score?"
Maybe the model is biased, how do we know? Is there an independent bias assessment process? Is there a way to sue for unfair rating? What are the values of the people working on the training set? Is there a "speak to human" step in the process where you can plead your case? I don't want an automated social discrimination machine.
Having to trade away your privacy for access to financial services (where this is headed) is not OK just because the spying is unbiased.
This is a common red herring, like the proliferating surveillance cameras and face recognition - somehow it is always implied the main reason we should worry is the racial bias, while everyone being followed 24/7 by a digital private investigator is a secondary concern.
Well said. I would also add that computing such a score requires the same kind of intensive surveillance that many people here on HN (including me) are morally opposed to.
Chinese social scoring system is much more extensive.
It includes jaywalking and speaking evil of the government.
It affects internet speed and boarding on trains.
Sure the “western countries” are heading that direction, but still have a long way
The blog is itself an advertisment for a longer (32 page) report released in August 2020, downloadable PDF available, largely an emerging trends report.
"Financial Intermediation and Technology: What’s Old, What’s New?"
The prospect of Fintech credit scoring from digital footprints is from "Onthe Rise of FinTechs – Credit Scoring using Digital Footprints", by Tobias Berg, Valentin Burg, Ana Gombović, Manju Puri, July 2018. 51pp.
Would information about CEO behavior be useful to investors? That's worth considering. Browsing history? Phone call patterns?
It's being worked on at the Harvard Business School.[1]
"We measure the behavior of 1,114 CEOs in six countries parsing granular CEO diary data through an unsupervised machine learning algorithm. The algorithm uncovers two distinct behavioral types: “leaders” and “managers”. Leaders focus on multi-function, high-level meetings,while managers focus on one-to-one meetings with core functions. Firms with leader CEOs are on average more productive, and this difference arises only after the CEO is hired. The data is consistent with horizontal differentiation of CEO behavioral types, and firm-CEO matching frictions. We estimate that 17% of sample CEOs are mismatched, and that mismatches areas associated with significant productivity losses."
I'm not a fan of the traditional credit reporting system. It is often a black box that makes it difficult to understand how the actual credit scoring algorithm works. Although one of the important aspects of the current system is that all the inputs are known. We don't know exactly to what degree each input matters, but we know the inputs. A new system needs to keep that aspect or else is going to cause pain for a lot of people who are financially punished without being aware of why. This type of transparency is also important for fairness. As other commenters have mentioned, some traits like race would be a good indicator of credit worthiness due to a variety of systemic socioeconomic issues. However that is generally considered unfair and is illegal to take that into consideration.
That is the content of a meme circulating recently which is imprecise at the most charitable (and IMO pretty close to entirely wrong)
Engineer Bill Fair and mathematician Earl Isaac (the “F” and “I” in FICO) started their company in 1956 and had a bank credit card scoring system available in 1970.
Equifax was founded in 1899 (not a typo). The original name: “Retail Credit Company”.
Set up a 'headless' chromium container with your social media cookies, set the user-agent to the very latest $5,000 Mac, and have it google stuff like "most exclusive golf clubs in xxx city" and "best winter tires to optimize insurance rates"
Few days of that and you'll qualify for a mcmansion mortgage in the monoculture part of the suburbs where all the dentists live.
Not so future ... some 6 years ago I have been advised by a recruiter that to apply to a certain SV company, I must have a Facebook profile with a minimum of 100 friends. Two days of playing some stupid Flash game got me 400 friends and a position at a company.
Readit News
Here is one of the most sensitive applications of ML. This one, like face recognition, can directly harm people.
These guys only seem to think about the "superiority" of their method in minimizing lending risk. But the other party into this deal can also have something to lose.
What's the difference between this and the Chinese social scoring system? They want to use private search logs. "Your privacy for a credit score?"
Maybe the model is biased, how do we know? Is there an independent bias assessment process? Is there a way to sue for unfair rating? What are the values of the people working on the training set? Is there a "speak to human" step in the process where you can plead your case? I don't want an automated social discrimination machine.
This is a common red herring, like the proliferating surveillance cameras and face recognition - somehow it is always implied the main reason we should worry is the racial bias, while everyone being followed 24/7 by a digital private investigator is a secondary concern.
There are multiple mutually incompatible metrics of fairness, being unbiased in one makes it biased in others. See e.g. https://www.chrisstucchio.com/pubs/slides/crunchconf_2018/sl...
Sure the “western countries” are heading that direction, but still have a long way
"Financial Intermediation and Technology: What’s Old, What’s New?"
https://www.imf.org/en/Publications/WP/Issues/2020/08/07/Fin...
The prospect of Fintech credit scoring from digital footprints is from "Onthe Rise of FinTechs – Credit Scoring using Digital Footprints", by Tobias Berg, Valentin Burg, Ana Gombović, Manju Puri, July 2018. 51pp.
https://www.fdic.gov/analysis/cfr/2018/wp2018/cfr-wp2018-04....
That was discussed at the time on HN (46 comments, June 22, 2018):
https://news.ycombinator.com/item?id=17377967
Chrome extension which changes your user agent request header to get you a better mortgage
Googling "Fun activities that don't cost money" increases your interest rates, study finds
Samsung partners with Equifax to offer better car loans to users who upgrade to a new phone
It's being worked on at the Harvard Business School.[1]
"We measure the behavior of 1,114 CEOs in six countries parsing granular CEO diary data through an unsupervised machine learning algorithm. The algorithm uncovers two distinct behavioral types: “leaders” and “managers”. Leaders focus on multi-function, high-level meetings,while managers focus on one-to-one meetings with core functions. Firms with leader CEOs are on average more productive, and this difference arises only after the CEO is hired. The data is consistent with horizontal differentiation of CEO behavioral types, and firm-CEO matching frictions. We estimate that 17% of sample CEOs are mismatched, and that mismatches areas associated with significant productivity losses."
[1] https://www.hbs.edu/ris/Publication%20Files/17-083_b62a7d71-...
Engineer Bill Fair and mathematician Earl Isaac (the “F” and “I” in FICO) started their company in 1956 and had a bank credit card scoring system available in 1970.
Equifax was founded in 1899 (not a typo). The original name: “Retail Credit Company”.
https://www.opploans.com/blog/a-brief-history-of-credit-scor...
https://www.myfico.com/consumer-division-of-FICO.aspx
Set up a 'headless' chromium container with your social media cookies, set the user-agent to the very latest $5,000 Mac, and have it google stuff like "most exclusive golf clubs in xxx city" and "best winter tires to optimize insurance rates"
Few days of that and you'll qualify for a mcmansion mortgage in the monoculture part of the suburbs where all the dentists live.
Future is bright!
I remember a uni network required me to install some anti-malware software. After changing my user agent, voilà, no requirement.
Take your data and shove it up your arse. If you work on this stuff, you're a cancer cell. You just haven't realised it yet.
I don't need your "credit". I don't even need you. No-one does.
But since you ask, there was "God View":
"'God View': Uber Allegedly Stalked Users For Party-Goers' Viewing Pleasure" (Kashmir Hill)
https://www.forbes.com/sites/kashmirhill/2014/10/03/god-view...