<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Matthew Chun on Medium]]></title>
        <description><![CDATA[Stories by Matthew Chun on Medium]]></description>
        <link>https://medium.com/@mattwchun?source=rss-cb0e2679a326------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*uMaLl-X6tx1CVcHHw9wRgQ.jpeg</url>
            <title>Stories by Matthew Chun on Medium</title>
            <link>https://medium.com/@mattwchun?source=rss-cb0e2679a326------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 01:14:57 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mattwchun/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Behavioral Models for Preferences]]></title>
            <link>https://medium.com/@mattwchun/behavioral-models-for-preferences-ef0b179cba4e?source=rss-cb0e2679a326------2</link>
            <guid isPermaLink="false">https://medium.com/p/ef0b179cba4e</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[behavioral-economics]]></category>
            <category><![CDATA[economics]]></category>
            <dc:creator><![CDATA[Matthew Chun]]></dc:creator>
            <pubDate>Fri, 23 Mar 2018 19:37:55 GMT</pubDate>
            <atom:updated>2018-03-30T20:42:53.832Z</atom:updated>
            <content:encoded><![CDATA[<p>Economists use utility to measure a person’s total satisfaction from consuming a good or service. In classical economics, we make <a href="http://www.econ.ucla.edu/sboard/teaching/econ11_09/econ11_09_lecture2.pdf">four assumptions</a> on preferences: completeness, transitivity, monotonicity, and convexity. Relaxing these assumptions and drawing from what we know from behavioral economics and psychology we can better model preferences in a number of ways.</p><h3>Prospect Theory</h3><p>Discovered by pioneers of behavioral economics Professor Daniel Kahneman and Professor Amos Tversky, prospect theory makes the following modifications and additions to the classical utility function:</p><ol><li>Losses hurt more than gains feel good known as loss aversion.</li><li>People make judgements based on a reference point not absolute values.</li><li>People display diminishing sensitivity. People feel good less and less as they gain more relative to their reference point. People hurt less and less as they lose more relative to their reference point.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/973/1*pdiz7_bnJoq3oYxJ0afLMw.png" /><figcaption>f(x, r) &lt; |g(y,r)| for all x, y when x- r = |y-r| and f’(x, r) &gt; 0, f’’(x, r) &lt; 0, g’(y, r) &lt; 0, g’’(y, r) &gt; 0</figcaption></figure><p>f(c,r)=sqrt(c-r) and g(c,r)=-3*sqrt(|c-r|) are examples of functions that would make the conditions hold. Graphically this looks like:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k7PL5AsAYOJnzPDR49xs_Q.png" /><figcaption>f(c,r) = sqrt(c-r) and g(c,r) = -3*sqrt(|c-r|)</figcaption></figure><p>Loss aversion is shown in the above graph by the greater magnitude utility for a loss than the utility from an equivalent gain. The reference point is incorporated in the utility function by setting the x axis to consumption minus reference point instead of just consumption. Lastly, diminishing sensitivity is demonstrated by the diminishing marginal utility (f’’(c,r) &lt; 0) in gains and diminishing magnitude marginal utility in losses (g’’(c,r) &gt; 0).</p><h3>Hyperbolic Discounting</h3><p>Hyperbolic discounting modifies exponential discounting because of people’s systematic behavior to heavily discount the near feature relative to the far future. Dynamic consistency is the property where the action a person thinks they will take in the future is the same as the action they take when the time comes. In hyperbolic discounting, dynamic consistency is no longer guaranteed to hold. The following formula is what a person under hyperbolic discounting aims to maximize at time t.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/810/1*eQSZpQlOi6JhXE5jV8IoRQ.png" /></figure><p>When beta is 1, we have the traditional exponential discounting formula. Otherwise, beta must be less than 1 and greater than or equal to 0 and signifies the short-term discount factor that underweights all the future periods.</p><p>This model allows us to study three types of thinkers: naive, sophisticated, and rational. When beta is one these three types of thinkers behave the same. However, when beta is less than one we can see how behaviors start to deviate.</p><h3>Anticipatory Utility</h3><p>In classical economics we typically assume people are rational and do not let their emotions bias their decisions. However, it is easy to find an example in the real world where emotions influence people to make irrational decisions. Anticipatory utility is one way to correct for these emotional biases. Anticipatory utility is the utility derived from setting certain expectations about the future. Similar to hyperbolic discounting, since we are modeling preferences based on the future, we use expected utility.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/986/1*El7apPjryj5ExqhAv7nkAQ.png" /><figcaption>f_A(p) is the anticipatory utility and p is the probability of the event the person expects. The sum is the expected utility formula where S is the set of all events and pi(s_i) is the probability of event s_i.</figcaption></figure><p>An interesting implication of anticipatory utility is it breaks an classical assumption that people always want to know as much information as possible. In this model and in the real world, people choose not to learn new information because the anticipatory utility could cause their overall expected utility to decrease. For example, anticipatory utility would explain why a student would avoid checking their final exam grade because they know it will make them unhappy.</p><h3>Projection Bias</h3><p>Another systematic bias people exhibit is projection bias. Projection bias is when people under-appreciate and underestimate changes in their preferences by falsely projecting their current preferences onto their future preferences. This bias is exhibited by hunger and grocery studies that show people under-appreciate the effects of their hunger state. This bias causes hungry people to buy more items not on their grocery list than currently satiated people (Gilbert, Gill and Wilson (1998)).</p><p>Projection bias is important to study because many people make poor decisions because they make the decisions in different states than what they trained in. This bias is the main motivator behind advice commonly heard in sports like Mike Candrea, Olympic coach, quote:</p><blockquote>“Practice like it’s a game; play the game like it’s practice”</blockquote><p>To model this bias mathematically we can loosely use a conditional utility notation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*twj86f56GppnamGRHZxiNw.png" /><figcaption>u(c_t, s_t) represents the utility from consuming c_t in the current state, s_t. The far right term, represents the projection bias.</figcaption></figure><p>The projection bias term can be defined as:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qhv5BSFtOGEaFL7e4nT3xQ.png" /><figcaption>Alpha is the degree of projection bias. If alpha is zero then there is correct understanding of future utility. The first term is the utility from consuming in the future c_{t+1} at the current state, s_t. The second term is the utility that one actually gets from consuming in the future c_{t+1} at the future state, s_{t+1}.</figcaption></figure><p>Projection bias can have major implications in fields like product design. Underestimating changes can bias product managers and designers to think others have the same needs and priorities as them. It is natural to think others are more similar to you than they actually are. This great <a href="https://blog.optimizely.com/2015/10/28/avoid-projection-bias/">article</a> explains how to navigate projection bias to build better products.</p><h3>Distributional Preferences</h3><p>The previous four models are for isolated situations where there is only one agent (person) in the game. However, there are many situations where there are more than one person in the <a href="http://www.econlib.org/library/Enc/GameTheory.html">economic game</a>. This and the next model “loosen” the classical assumption that people are only self-interested. People are willing to hurt themselves to make a statement. An example is people are willing to spend more money on products made by local firms to show their support for their community and small businesses.</p><p>The first way to incorporate selfless behavior into the utility model is to use distributional preferences. Distributional preferences are preferences where utility is represented in terms of their own material resource payoff and others’ material resource payoffs. To simplify distributional preferences mathematically, we set the number of the agents in the game to be two so the utility for player 1 looks like:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s0R7V4eUXi--bYm9ngcSiA.png" /><figcaption>The utility for player 1. Rho represents how altruistic player 1 is when they are ahead and sigma represents how altruistic player 1 is when they are behind.</figcaption></figure><p>Rho is typically larger than sigma because people tend to be more selfless when they are ahead. In other words, it’s easier to be nice when you win than when you lose. It is possible for sigma to be less than zero. When sigma is zero, it means player 1 is very unhappy when player 2 is ahead.</p><h3>Rabin’s Fairness Model</h3><p>Distributional preferences represent utility only in terms of the amount of money others get. However, the process to the end payoffs for all the players in the game also matters. This is the motivation behind the intentions-based preference model that splits utility of a person into two components: material payoffs and psychological payoffs. Professor Matthew Rabin developed a fairness model that can explain behavior in the three possible cases:</p><ol><li>Player 1 gets additional psychological utility for being nice if they think Player 2 is being nice to them.</li><li>Player 2 gets additional psychological utility for being mean if they think Player 2 is being mean to them.</li><li>Player 1 will just maximize material payoffs (same as the distributional preferences model) if they think Player 2 is being neutral.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/739/1*VhustrwnCgQN1nyf1sJZMw.png" /><figcaption>The utility for player i. k_{ij} is the kindness of player i to player j. Alpha is the degree of reciprocity of player i. x_i is the material payoff for player i.</figcaption></figure><p>We can think about the product in the brackets as positive when both of the players are kind or when both of the players are not kind. When one is kind and the other is not kind, then the product is negative. This model is highly extendable because the kindness terms can be recursive. They can be defined as a function of what the other player has done in the past as well as what they think they will do next.</p><p>Each of these models are used to study irrational behavior. The application of these models are immense as they can be applied to recommender systems in the video streaming and e-commerce domains. It would be interesting to incorporate these biases into algorithms and test their effectiveness. I only listed six but there are a number of other biases as seen by this expansive wikipedia list:</p><p><a href="https://en.wikipedia.org/wiki/List_of_cognitive_biases">List of cognitive biases - Wikipedia</a></p><p>It would be great to hear how you might combine these six models. The naive method is to assign a weight to each of the models and sum them up.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ef0b179cba4e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Insta-Message Project (Nexus UW, Colourhood, Hearth Seattle)]]></title>
            <link>https://medium.com/@mattwchun/insta-message-project-nexus-uw-colourhood-hearth-seattle-966b020e656f?source=rss-cb0e2679a326------2</link>
            <guid isPermaLink="false">https://medium.com/p/966b020e656f</guid>
            <category><![CDATA[software]]></category>
            <category><![CDATA[nexus-uw]]></category>
            <category><![CDATA[mobile-app-development]]></category>
            <category><![CDATA[social-work]]></category>
            <category><![CDATA[nonprofit]]></category>
            <dc:creator><![CDATA[Matthew Chun]]></dc:creator>
            <pubDate>Fri, 16 Mar 2018 18:40:18 GMT</pubDate>
            <atom:updated>2018-03-16T18:40:18.251Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AmD_dhlZZ4txcBkl5tSXTA.jpeg" /><figcaption>The Insta-Message project team with Hearth Seattle’s founder, Robert Snow</figcaption></figure><p>Today, I am happy to announce <a href="https://www.facebook.com/NexusUW/">Nexus UW’s</a> inaugural project. The mission of Nexus UW, a University of Washington club, is to help those in society that cannot afford to develop software by connecting them with students eager to apply their skills they learned in the classroom.</p><h3><strong>Project Description</strong></h3><p>The goal of the Insta-Message project’s is to build a mobile messaging framework and mobile application to help Hearth Seattle, a nonprofit organization that helps displaced Seattleites, by assisting them and social workers with the referral process. A team of nine students and two experienced professionals with varying but typical roles you would see in a software development team will develop the project over the <a href="http://www.washington.edu/">UW</a> Spring academic quarter (~10 weeks).</p><h3><strong>Why is this important</strong></h3><p>The founder of Hearth Seattle, Robert Snow, an experienced and passionate social worker uses an example he experienced to explain the project’s importance. The past Sunday he was notified that a sex worker in Seattle needed a safe place for the night and didn’t want to call the authorities because she did not want to get involved with the authorities, which is common in social work. Mr. Snow ended up giving her a referral to a shelter via a phone call. This may seem like a simple and flawless process, but as many social workers will tell you there are many problems with this process. The main three problems are:</p><ol><li><strong>This will happen again in the future </strong>— Mr. Snow explained that most of his job as a social worker is related to this referral process. He finds that the same people come back to him with the same questions because they forgot or lost the information they need to use the resources he provided. The mobile application will give those in need a documented history of questions and answers.</li><li><strong>The communication method is far from perfect for those in need</strong> — So, why not use Facebook messenger or one the other thousand messenger apps? One of the main reasons is from Mr. Snow’s experience many of the people he helps have a distrust for these applications. An advantage of this application is it will give users quick and direct access to social workers who know immediately why they are contacting them.</li><li><strong>The communication method is far from perfect for social workers </strong>— This is the main reason for not using apps like Facebook messenger because many messaging applications are linked to social network accounts. Mr. Snow and many other social workers do not want to mix their personal and work lives. So, a separate application will allow social workers to clearly draw this line.</li></ol><p>The main problem this project is solving is the social worker’s referral process. It is incredibly important as many shelters require a referral. Therefore, a strong, secure, and confidential communication method makes giving out a referral much easier. The majority of the community that will use the messaging application do in fact have smartphones. This means the application will be accessible to all as the only requirement is an internet connected device. The application will give those in need a central place where they can reach out to social workers and quickly find the resources they need to make their lives a little easier.</p><h3><strong>The Team</strong></h3><p><strong>Project Managers</strong></p><p>Leading the Insta-Message project as the project managers are Andrei Villasana, the founder of Colourhood, and Katherine Li. Both are experienced Software Engineers and will oversee the development and design of the Insta-Message project.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JO_KHBMvlsOqv40-VqYU0A.png" /><figcaption><strong>Project Managers</strong>: Andrei Villasana (left) and Katherine Li (right)</figcaption></figure><p><strong>Developers</strong></p><p>The development team range from Sophomores to Seniors that major in <a href="https://www.cs.washington.edu/">computer science</a>, <a href="https://ischool.uw.edu/">informatics</a>, or <a href="https://econ.washington.edu/">economics</a>. This team of five students will be responsible for developing the iOS application with Swift, Android application with Kotlin, and Backend with NodeJS.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XeLJt9c5zVqvy5mi7o_hZw.png" /><figcaption><strong>Developers</strong>: Quan Tran, Frank Zhu, Jong Ho Lee, Jacob Lu, Richard Jiang (top left to bottom right)</figcaption></figure><p><strong>Designers</strong></p><p>The design team consists of four designers that also range in class standing with diverse backgrounds from the <a href="https://www.hcde.washington.edu/">Human Centered Design &amp; Engineering</a> to <a href="http://larch.be.uw.edu/">Landscape Architecture</a> departments at the University of Washington. They will be responsible for sketching, wire-framing, and much more to develop an incredible user interface and experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZJ6j86C3qpJ_z5xXNKJBeA.png" /><figcaption><strong>Designers</strong>: Inhae Kim, Rachel Ren, Jae Jung Park, Monica Posluszny (left to right)</figcaption></figure><h3>Future</h3><p>Nexus UW is happy to have such a talented team of project managers, developers, and designers working on the Insta-Message project. More updates will follow!</p><p>If you’re interested in working on projects like this in the future, we at <a href="https://www.facebook.com/NexusUW/">Nexus UW</a> host at least one project like this for a Seattle non-profit per academic quarter. Like us on <a href="https://www.facebook.com/NexusUW/">Facebook</a> to get updates and reach out at <a href="/service/mailto:uw.nexus@gmail.com">uw.nexus@gmail.com</a> to learn more!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=966b020e656f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>