Homelessness is an unsolved problem. Most people would agree that no one should go without basic needs in a high-income, developed country (i.e. Maslow’s Hierarchy of Needs tier 1 and 2 (Maslow 1943)). Yet a 2017 point-in-time study estimated there were 553,000 people experiencing homelessness on a on a single night in the US, a year marking the first rise in homelessness in 7 years(Henry et al. 2017). And roughly 1 in 3 of those persons were living in unsheltered locations (Henry et al. 2017). More concerning, an estimated 2.5 million children are homeless each year in the US (Bassuk et al. 2014). And San Francisco City estimates there were 7,499 homeless persons in 2017 (San Francisco City 2017), many of them with full-time employment (Har 2017). Chronic homelessness costs the US an estimated $10 billion annually in public funds (Oxford Analytica 2006) and leads to increased morbidity and mortality (Fazel et al. 2014).
Technology is an inexpensive and scalable solution to many problems. However, the causes of homelessness are complex (Brown et al. 2016; Stanhope and Dunn 2011) and so are sustainable solutions (Sheltr 2018). Serious concerns have been raised about using software to help disadvantaged populations. People experiencing homelessness may not have been represented in the data used to create computer algorithms (inclusion)(Hong et al. 2017; Buolamwini and Gebru 2018; Jha et al. 2009), and may be harmed when computer-aided decision-making is used to allocate funding of basic needs (equity) (Eubanks 2018).
While many people in large cities across the United States encounter panhandlers on their morning commute, the impulse to give may be disincentivized. People have expressed fear that giving cash might not solve the problem (Bulman 2018) or may support drug economies (DeBeck et al. 2007; Bose and Hwang 2002). While panhandling is often discouraged, evidence shows that some income from panhandling does go to help the basic needs of the homeless (Bose and Hwang 2002), but increasingly adults may not carry cash for donations (in 2017, up to 50% of Americans (U.S. Bank 2017)). For those who do, the interaction and need is unrecorded, and not represented in public datasets(Bose and Hwang 2002).
A person experiencing homelessness is unlikely to have a cell phone and is therefore excluded from many datasets used to create artificial intelligence (AI) algorithms (Leidig and Teeuw 2015). However a person who does not carry cash (62% <age 35 ) (Wang 2015) is very likely to have a cell phone (95% < age 30 (Anon 2018). Simple apps like Shazam (Shazam Entertainment Limited 2018) have been used to successfully capture impulse purchases for music sales. The desire to give a panhandler $1 might be considered an impulse purchase of homeless services (Taute and McQuitty 2004; Gaetz and O’Grady 2002) which could be captured more effectively with a simple mobile app (Dordick and O’Flaherty 2017). Capturing this information at scale could provide a digital voice for people experiencing homelessness that was previously lacking in public datasets.
A natural experiment for this theory exists among three local towns. Recently the first town’s city council invested in a shelter and long-term assistance program for homelessness, while the neighboring towns did not invest in the same level of services. Advocacy groups in the first city are concerned that the current shelter is not large enough to serve the population of homeless within the city. The town council is concerned that people have migrated from the neighboring towns, following the services and overwhelming local resources . There is very little data available to resolve the debate which advocates continue to address in town council meetings on behalf of the homeless population of the area.
Theoretically, an app that facilitated the impulse purchase of homeless services between someone owning a smartphone in the area and local homelessness advocacy groups could provide a digital voice for the unhoused population at policy debates, foster behavior change to create new donors among young people, and sustainably redirect empathic donations to structured human-led solutions. Pairing the app with formal research from public sources and local stakeholders (advocates, community leaders, and people experiencing homelessness) could generate a more inclusive and equitable dataset for predictive algorithms.
There is benefit from integrating snapshot assessments of need in policy decisions. It is true that “slow-thinking” humans may be better positioned to create lasting benefit and ensure equity from donations to structured homeless services (Sheltr 2018; Kraus and Keltner 2009), However “fast-thinking” thin-sliced, spot assessments of socioeconomic status can be accurate and constitute a underutilized datapoint in homelessness needs (10, 29),. Additionally “fast-thinking” intuitive technology may be able to improve on cash donations to equitably distribute funds to these organizations while highlighting population/funding disparities for policymakers (23, 24).
It is likely homelessness is an algorithm a computer can never solve, but a computer can help humans fund humans to solve it (Marrow 2012).
Anon 2018. Demographics of Mobile Device Ownership and Adoption in the United States [Online]. Available at: http://www.pewinternet.org/fact-sheet/mobile/ [Accessed: 30 September 2018].
Bassuk, E.L., DeCandia, C.J., Beach, C.A. and Berman, F. 2014. America’s youngest outcasts: A report card on child homelessness.
Baum, A., Scarpa, J., Bruzelius, E., Tamler, R., Basu, S. and Faghmous, J. 2017. Targeting weight loss interventions to reduce cardiovascular complications of type 2 diabetes: a machine learning-based post-hoc analysis of heterogeneous treatment effects in the Look AHEAD trial. The lancet. Diabetes & endocrinology 5(10), pp. 808–815.
Bose, R. and Hwang, S.W. 2002. Income and spending patterns among panhandlers. Canadian Medical Association Journal 167(5), pp. 477–479.
Brown, R.T., Goodman, L., Guzman, D., Tieu, L., Ponath, C. and Kushel, M.B. 2016. Pathways to Homelessness among Older Homeless Adults: Results from the HOPE HOME Study. Plos One 11(5), p. e0155065.
Bulman, M. 2018. Should we give homeless people money? | The Independent [Online]. Available at: https://www.independent.co.uk/news/uk/home-news/should-we-give-homeless-money-a8124951.html [Accessed: 28 September 2018].
Buolamwini, J. and Gebru, T. 2018. Gender shades: Intersectional accuracy disparities in commercial gender classification. Conference on Fairness, Accountability and Transparency, p. 77.
DeBeck, K., Shannon, K., Wood, E., Li, K., Montaner, J. and Kerr, T. 2007. Income generating activities of people who inject drugs. Drug and Alcohol Dependence 91(1), pp. 50–56.
Dordick, G. and O’Flaherty, B.A. 2017. Policy for Panhandling: How to Encourage Good Panhandling and Discourage Bad.
Eubanks, V. 2018. Automating inequality: How high-tech tools profile, police, and punish the poor. St. Martin’s Press.
Fazel, S., Geddes, J.R. and Kushel, M. 2014. The health of homeless people in high-income countries: descriptive epidemiology, health consequences, and clinical and policy recommendations. The Lancet 384(9953), pp. 1529–1540.
Gaetz, S. and O’Grady, B. 2002. Making Money. Work, employment and society 16(3), pp. 433–456.
Har, J. 2017. “We Still Need to Eat”: Tech Boom Creates Working Homeless | California News [Online]. Available at: https://www.usnews.com/news/best-states/california/articles/2017-11-07/in-shadow-of-tech-boom-the-working-homeless-sleep-in-cars [Accessed: 28 September 2018].
Henry, M., Watt, R., Rosenthal, L. and Shivji, A. 2017. The 2017 annual homeless assessment report (AHAR) to Congress. Washington: The US Department of Housing and Urban Development.
Hong, Y.A., Zhou, Z., Fang, Y. and Shi, L. 2017. The digital divide and health disparities in china: evidence from a national survey and policy implications. Journal of Medical Internet Research 19(9), p. e317.
Jha, A.K., DesRoches, C.M., Shields, A.E., Miralles, P.D., Zheng, J., Rosenbaum, S. and Campbell, E.G. 2009. Evidence of an emerging digital divide among hospitals that care for the poor. Health Affairs (Project Hope) 28(6), pp. w1160-70.
Kraus, M.W. and Keltner, D. 2009. Signs of socioeconomic status: a thin-slicing approach. Psychological Science 20(1), pp. 99–106.
Leidig, M. and Teeuw, R.M. 2015. Correction: quantifying and mapping global data poverty. Plos One 10(12), p. e0145591.
Marrow, J. 2012. 16 Top Nonprofits Fighting Bay Area Homelessness Named by 100 Experts [Online]. Available at: https://trust.guidestar.org/blog/2012/11/09/16-top-nonprofits-fighting-bay-area-homelessness-named-by-100-experts/ [Accessed: 30 September 2018].
Maslow, A.H. 1943. A theory of human motivation. Psychological Review 50(4), pp. 370–396.
Oxford Analytica 2006. Cutting The Cost Of Homelessness In U.S. [Online]. Available at: https://www.forbes.com/2006/08/25/us-homeless-aid-cx_np_0828oxford.html [Accessed: 27 September 2018].
San Francisco City 2017. Homeless Point-In-Time Count And Survey 2017: Executive Summary.
Shazam Entertainment Limited 2018. Shazam [Online]. Available at: https://www.shazam.com/ [Accessed: 28 September 2018].
Sheltr 2018. SHELTR [Online]. Available at: https://www.hackinghomelessness.com/ [Accessed: 28 September 2018].
Stanhope, V. and Dunn, K. 2011. The curious case of Housing First: the limits of evidence based policy. International Journal of Law and Psychiatry 34(4), pp. 275–282.
Stuckler, D. and Basu, S. 2013. The body economic: why austerity kills. Basic Books (AZ).
Taute, H. and McQuitty, S. 2004. Feeling good! doing good! an exploratory look at the impulsive purchase of the social good. Journal of Marketing Theory and Practice 12(2), pp. 16–27.
U.S. Bank 2017. U.S. Bank Cash Behavior Survey [Online]. Available at: https://www.usbank.com/newsroom/news/digital-payment-platforms-primed-to-topple-cash.html [Accessed: 27 September 2018].
Wang, C. 2015. Cash | Who Holds Cash? Evidence from the 2012 Diary of Consumer Payment Choice [Online]. Available at: https://www.frbsf.org/cash/publications/fed-notes/2015/december/who-holds-cash/ [Accessed: 30 September 2018].