The prototype: applying persuasive technologies to influence regenerative behavior

I’ve begun PART 1 of this series of articles stating that the root cause of our metacrisis lies in our inability to see ourselves as interconnected parts of the whole. In PART 2, I shared a dialogue that explains much of my political-economic background. PART 3 presents a theoretical scope of the proposed solution, a wholeness-based global collective intelligence tool. Below, I introduce a research proposal for the prototype implementation and explain how it leverages the potential of persuasive technologies to support and scale regenerative behavioral change.

Introduction

Anyone with access to global news is probably overwhelmed by the chaos that piles up exponentially. While many may feel hopeless about the horrors of what appears to be the last stage of a dying system, I feel blessed to at least have visualizations of alternative pathways we can follow to leave this mess and journey into a brighter future. Thankfully, many people are actively working for systematic social change, and as part of the group, I'm here to deliver words of hope and an action plan.    

As a refresher, click the dropdown below for a recap of the extended vision provided in PART 3.  

To help you visualize the proposed solution and address its complexity, I hereby present a research proposal and pilot program to test its main hypothesis. The pilot program is divided into 3 phases: prototype development, community activation, and research publication. Each phase has a dedicated section in this text covering the following elements:

• Phase 1: Prototype Development - explores the use of behavioral science and persuasive techniques to justify the prototype implementation and expand on its features and possible effectiveness;
• Phase 2: Community Activation - shares the plans for community activation in partnership with an educational NGO at a favela in Rio de Janeiro, with pieces of storytelling to introduce my dear friend Carol, the NGO founder;
• Phase 3: Research Publication - describes how to measure the prototype's impact and the importance of disclosing the findings in a research publication, along with open-sourcing the code and other outcomes.

Before we dive into the specifics, let's clarify some structural ideas behind this peculiar task management tool. Although the application has many layers, its primary purpose is to support and scale regenerative behavioral change. This means it is designed to guide decision-making in favor of actions that are aligned with individual, collective, and ecosystemic goals and long-term benefits. As a behavioral research object, the intended outcome can be translated as an increase in people's routines of actions related to mutual aid, communal participation, and environmental stewardship. To achieve such change, let's follow the hypothesis that:

We all have intrinsic forces guiding our desires, choices and behaviors. A proper information system can use this knowledge for matchmaking and increase mutual aid, communal participation, and/or environmental stewardship, regardless of financial incentives. More poetically speaking, everybody cares about something, once we know the drives of love and caring intrinsic to us we can (and should) use them as primordial forces guiding human labor and organization.

The trickiest part of testing this hypothesis is understanding what the proper information system looks like. What kind of information should we exchange to help us operate in that sweet spot in which what's good for me is also good for the whole, and vice versa? To answer that, I propose the development of a personalized task recommendation engine as the key component of the prototype. As the name suggests, it's a tool to recommend tasks and behaviors based on individual inputs. That's how we plan to connect the inner drives of love and caring to the practical demands for human labor and organization.

The backbone of this engine is an action-oriented tag system that prioritizes regenerative actions. This list of actions and objects will be used to train machine learning models to facilitate matchmaking, plus it will help us organize toward the same goal of acting for the well-being of the whole.    

Having a pre-established Global Shared Goal and making its agreement a mandatory process to access the prototype indicates, right at the gate, the purpose of this cyberspace. Every user must agree TO NURTURE LIFE to join this infinite game. This is an attempt to unite participants (humanity) under the same objective, one that is broad and ambiguous enough to accommodate all our differences and conflicts of interest, but still compelling to the tasks we must hold as part of the living system, such as actions of self-care, humanitarian service, and ecosystemic stewardship.

Once the global shared goal is set and we know all people playing the game agreed to work towards the same objective, the nurturing of ANY life, we can coordinate around that and build a global collective intelligence (GCI) tool to exchange information and enhance our self-organizing abilities. How do you choose to nurture yourself and those around you? This is what the personalized task recommendation engine seeks to know, track, and support.  

A comprehensive list of predefined tags can be used to organize information in a task-oriented language that highlights the ACTION + OBJECT at stake. Each tag, representing an action, location, resource, or living species, becomes a folder where all associated data can be globally stored and compiled, facilitating data analysis and visualization. With the assistance of AI models and search/filtering mechanisms, participants can harvest and process the public data they have created to use it for social studies and informed decision-making.

Let's say you want to nurture life by 'feeding Palestinians', and you set it as a topic of interest. Your task recommendation engine would take a look at this action tag, search for people and groups around you who are publicly working on it, and show task options to get you involved. Under a tag, people can share and find all sorts of relevant data: offers, requests, media, polls, metrics, discussions, reviews, people, nearby activity, guides, resources, research, etc. The same goes for tasks and their specific combination of tags. For instance, the subtasks of 'siege-breaking Gaza' and 'protesting Israel' can be assigned as tags to compile the data created by those working on them.

The tag system and the task-oriented language serve as a foundation for the personalized task recommendation engine and the dynamic workflow structure in which the subtasks of nurturing life can be registered and analyzed.

Following this task structure, people can share what they've done and coordinate the to-do list in a way that is not siloed, after all, each task is a global reservoir designed to help people exchange notes on their (nurturing) activities. Here, cooperation is the standard, and to further assist self-organizing habits and help us deal with violations, we can apply commons management techniques, such as setting boundaries, conflict mediation, and implementing peer-to-peer sanctions.  

The hope is that an information system like this, centered on actions and prompted by unity and stewardship, can reconnect us to what truly matters. Nowadays most of us are so desperately trying to make it that we let the hustle culture blind us to the truth... we have already made it! We are alive!!! From all the odds, we have been gifted existence, and now everything we do leaves a mark on the fabric of reality. That's what matters, how we choose to use our autonomy and the extreme power of creation (and destruction) granted to us.

There's so much one person can do, so many beings they influence throughout their life, and yet being 1 part in 8 billion makes us judge ourselves and others as insignificant, even though we are unique, with a singular set of skills, desires, knowledge, and connections. Let's not fall for the paradoxical tricks of nature. As human beings, our power of creation is the most absurd, especially when we act as a collective.

If we wish to master this god-like power, we must bring more intention and consciousness into our actions and decision-making process. For that, we must exchange truthful information about our experiences, plans, and activities, so we can discuss the impact of our choices and decide the best goals for the future accordingly.

Therefore, a platform that integrates individual and collective actions and allows the simplest tasks to be shared, stored, and evaluated can help with that by bringing awareness to what we choose to do and provoking people to express, measure, and display the impact of their existence.

Can we trick people into acting for the well-being of the whole? Can we persuade them to become more supportive of their peers and environment? Now that you have had a glimpse into the proposed information system and its purpose, let's answer these questions by dissecting the prototype and its main features.

Phase 1: Prototype Development &  Persuasive Technologies

“We've used the same psychological techniques that apps like Instagram, TikTok or mobile games use to keep people engaged, but in this case, we use them to keep people engaged but with education… If we want to get people to do something meaningful, you can use the same techniques that apps like social media use to get people to do it. And even if you're not as engaging as those apps are, you can still get hundreds of millions of people to use your product.”  

Duolingo founder, Luis Von Ahn in the TED talk How to Make Learning as Addictive as Social Media

In this quote, the Duolingo founder is talking about persuasive technologies, a field in behavioral science that gained popularity with B.J. Fogg's book Persuasive Technology: Using Computers to Change What We Think and Do (2003). In it, the Stanford professor presents several persuasive principles and uses the acronym Captology (Computers as Persuasive Technologies) to name the domain of research, design, and application of human-computer interactions that explicitly intend to change, shape, or reinforce people's attitudes and behaviors without using coercion or deception. [1]

Even if you have never heard of captology, you have probably seen its usage in many different digital products. Rating systems, notifications/reminders, praises, rewards, scores, and testimonials are a few examples of persuasive techniques we see everywhere, from video games to fitness and wellness apps. Unfortunately, educational and health-targeting behaviors, such as learning languages, smoking cessation, dieting, and exercising, are not the most popular use cases of captology we have been exposed to. Rather, Big Tech companies take that title by exposing billions of people to machine learning algorithms that learn people’s behavioral patterns just so they can hit us with tailored advertisements.

Additionally, these companies have been using our best understanding of how the brain works to get us hooked on their products, thus giving them the power to shape users’ information and choice environments and making their algorithms indispensable for navigating the vast digital landscape. However, the relationship between these platforms and people is profoundly asymmetric: they have deep knowledge of users’ behavior, whereas users know little about how their data is collected, how it is exploited for commercial or political purposes, or how it and the data of others are used to shape their online experience. [2]

These asymmetries in Big Tech’s business model have created an opaque information ecology that undermines not only user autonomy but also the transparent exchange on which democratic societies are built. With that, we see persuasive technologies associated with several problematic social phenomena, such as the spread of false information, which includes disinformation (intentionally fabricated falsehoods) and misinformation (falsehoods created without intent, for example, poorly researched content or biased reporting), or attitudinal and emotional polarization. [2]    

Needless to say, captology is a controversial field of study. But what if we took the chapter on ethical concerns seriously and made the designer's intentions, methods, and outcomes transparent to the users? [1] What if we use it to support and scale regenerative behavioral change? The widespread use of Big Tech's platforms is indicative of the huge potential of persuasive technologies. We should take advantage of such knowledge to support mutual aid and environmental stewardship as an opposition statement to the tech giants and their continuous use of persuasive techniques for harmful profit-seeking purposes.  

"The way you overcome a powerful motivation is with another powerful motivation. If we can get people to fall in love again with being — within, between, and beyond — we can break the game-theoretic."

John Vervaeke in the conversation The Psychological Drivers of the Metacrisis

Researchers Oinas-Kukkonen and Harjumaa (2009) proposed seven postulates to be addressed when designing or evaluating persuasive systems. The first one helps understand the importance of such systems and states that information technology is never neutral. It is 'always on', influencing people’s attitudes and behavior in one way or another. Moreover, people are constantly being persuaded in a manner similar to how teachers persuade students in schools, and there is nothing wrong with it in itself. [3]

My take is that persuasion is an inherent part of human communication and social dynamics. It got naturally translated into software language, and as computers became more interactive and powerful, so did their ability to influence people.

Following the development of the Internet, smartphones, and wearable devices, individuals now generate a myriad of big data, much like walking data factories. Persuasion strategies powered by individuals’ big data are more advanced than previous ones. This increases the probability of successful interventions toward changes in users’ attitudes and behaviors. [4]

Whether we are conscious of it or not, we are constantly being persuaded. The modern habit (sometimes addiction) of sharing our lives online and creating huge digital footprints has become a behavioral goldmine for tech companies, and sadly, it is mostly being exploited for corporate greed. People should have ownership and the agency to mine their data treasure to use it for personal and collective benefit.

So let's present them with an ethical alternative. An information system that applies data-centered persuasion might have the power to support massive behavioral change; this type of cultural change, which affects both individual and collective levels, is what we need to reorganize human labor and hopefully escape the metacrisis. Instead of falling victim to the usual pushed behaviors of doomscrolling, voting for liars, arguing with bots, and buying things we don't need, we can opt for a persuasive system that motivate us to act toward what we care about, help us self-organize in local units, inspire us to recreate and have soverignty in our life-supporting systems, and ultimatly avoid the exploitative and destructive patterns imposed by capitalism.

Persuading a user is a multi-phased and complex task. [3] It is not easy, but with your support, I hope to try it out. To expand on how, here are a few software design ideas with their respective persuasive techniques.  

Personalized task recommendation engine

Recommendation systems (RS) are algorithms based on artificial intelligence, mostly on machine learning techniques, that support user-tailored decision-making by providing suggestions out of a wider catalog based on the users’ or like-minded users’ past choices or personal information. On the one hand, by making effective recommendations, RSs may help users navigate the environment by alleviating choice overload and decision fatigue. Indeed, even without invoking the highly disputed paradox of choice, users equipped with bounded rationality and limited cognitive resources may find it easier to make decisions aligned with their interests and goals when irrelevant options are filtered out. On the other hand, in virtue of their reach of impact and influence, RSs may reshape, to a large extent, the media landscape and the social dimension in which users interact and, consequently, affect the way people think and choose. [5]

RSs have such a behavioral impact because they usually follow two captology principles: tailoring and suggestion. The tailoring principle implies persuasion through customization, stating that information provided by computing technology will be more persuasive if it is tailored to the individual’s needs, interests, personality, usage context, or other factors relevant to the individual. The timely suggestion principle implies intervention at the right time and states that a computing technology will have greater persuasive power if it offers suggestions at opportune moments. [1]

Every time we open a social media app, we experience these principles in action as we scroll through new content and advertisements. Because of the secrecy behind these RSs, we are often unaware of how the algorithm is tracking and monitoring individuals' data, who companies are sharing our data with, and their intentions in affecting our worldview and decision-making. [5] If we wish to avoid such misconduct and build an ethical alternative, the first step is to provide transparency to the mechanisms behind our personalized task recommendation engine.

As mentioned in the introduction, our primary goal is to support and scale regenerative behavioral change, which involves increased actions of mutual aid, communal participation, and/or environmental stewardship. Taking care of ourselves, our communities, and the ecosystem surrounding us, these are the target behaviors we intend to achieve through data-centered persuasion.

To implement the tailoring and suggestion principles, we must collect individuals’ behavioral data, the more the better. Interests, skills, opinions, location, calendar, and most importantly, the nurturing tasks they want to do and the nurturing tasks they need done. This information, which should be protected under privacy protocols, will serve as the baseline for the recommendation system to motivate people to self-organize in two ways: by creating a newsfeed with tailored peer-uploaded tasks and by directly pinning people with tasks that match their profile.

From the start, users are asked to nurture life as the ultimate call to action; they can answer to it by uploading tasks from three different categories: personal, offer, or request. Personal tasks are private and should follow self-set goals and self-care routines, such as studying or dieting; offers are provision tasks done by the individual for others; and requests are appropriation tasks the user wants to find someone else to do. To avoid free-riders and those only interested in taking, a graph of completed tasks showing the balance between the 3 types can be displayed in people's profiles.    

To function well, the personalized task recommendation engine must know how you want to contribute to society and when you are most available to do it. If it's successful in providing you with irresistible matching task options, it's applying a third persuasive principle. The reduction principle implies persuasion through simplification and states that using computing technology to reduce complex behavior to simple tasks increases the benefit/cost ratio of it and influences users to perform the behavior. [1]

If I ask you to find a volunteer position right now, how easily and quickly would you find one that looks fun and aligned with your interests and agenda? I assume it would require a bit of effort. Therefore, to increase the likelihood of communal participation, we must reduce the effort.

Overall, the math is simple: if we wish to persuade people into self-organizing, we must know what they are intrinsically into, track when they are up for it, then make a matching task pop up in front of their faces, making participation as easy as possible. That's what the personalized task recommendation engine will be designed to achieve.    

However, it's worth noting that individuals may often have an unrealistic or inflated view of their preferences and personality due to a general lack of introspective skills and the tendency to self-deception. In such cases, RSs may provide a more clear-sighted representation of the user's psychological profile, thus supporting descriptive autonomy by providing reliable and transparent feedback on their objective tastes and needs and helping people realign their choice behavior. [5] To support such a claim, we can apply more persuasive techniques and enhance the software design with the following features.

Inquiry with Capy Vera

Besides the fact that some people may not have clarity about their interests or consolidated ideas for projects and tasks within their communities, it is also important to consider that even if they do, they might not be willing to use our product to develop them.  

Most of the time, we’re not actually “choosing” what to do next, at least, we’re not choosing consciously. Most of our daily behavior is governed by our intuitive mode. We’re acting on habit (learned patterns of behavior), on gut instinct (blazingly fast evaluations of a situation based on our past experiences), or on simple rules of thumb (cognitive shortcuts or heuristics built into our mental machinery). Researchers estimate that roughly half of our daily lives are spent executing habits and other intuitive behaviors, and not consciously thinking about what we’re doing. [6]

Therefore, using our product and incorporating it into people's lives is a behavior we must persuade them to adopt in itself; otherwise, it will be hardly used because people are busy with their intuitive behaviors. B. J. Fogg developed the Behavior Model (B=MAT), saying behavior occurs only when three elements converge at the same moment: motivation, ability, and a trigger. We'll discuss the motivating factors to use our product in Phase 2: community activation. For now, let's consider that our target audience has the ability to use smartphones and focus on the triggering elements.

For the personalized task recommendation engine to work, we must ensure that a trigger will prompt the behavior of uploading tasks. Instead of passively waiting for people to do it, we can take a more persistent approach and use AI agents to ask people what we need to know and gather more data in an open conversation.

For instance, every night we can provide a cue to journal about the activities of the day. With the answer, the language processing agent can then identify the tasks and automatically suggest that they get included in the user's profile with one simple click, helping the recommendation system to understand users' skills and routine, therefore enhancing matchmaking effectiveness.  The same goes for understanding individuals' goals for themselves and their communities. We can ask what improvements they wish to see, and with the response, the agent can suggest project ideas and sustainable local practices that could be implemented with detailed action plans, tasks, and needed resources.

The strategy of bypassing the conscious effort to upload a task on an unfamiliar app and replacing it with a prompt for conversation can be especially useful for people with limited ability to handle software, and people searching for connection, as the motivation to act (replying) will be stronger. Which leads to another persuasive potential of computers, their ability to play the role of social actors.  

The choice of having a capybara mascot named Vera has a persuasive purpose. People can respond to computers as though they were living creatures, even getting emotionally involved with them. This allows, for example, people to feel motivated by praise from the computers and a willingness to repay favors the computers have done for them. When computing products adopt the role of a social actor, they persuade people by applying the same persuasion principles that humans use to influence others. For instance, as social actors, computers can persuade people to change their attitudes or behaviors by rewarding them with positive feedback, modeling a target behavior, or providing social support. [1]

Hence, adding a mascot to the software design is worth the effort as it provides the opportunity to explore different types of social cues and influence, as peer pressure, social comparison, and facilitation. For example, if the mascot is cute enough, it can follow the attractiveness principle, which says that a computing technology that is visually attractive to target users is likely to be more persuasive. Also, a mascot allows us to experiment with interactive physical cues, like movements and eyes/facial expressions, which evoke feelings and emotions and might enhance the effect of the praise principle, that states that by offering praise, via words, images, symbols, or sounds, computing technology can lead users to be more open to persuasion. [1]

Interestingly, the relationship between inquiry and computers playing the role of social actors was one of the first phenomena observed during the advances of artificial intelligence, more specifically, the development of natural language processing. In the 1960s, MIT researcher Joseph Weizenbaum created ELIZA, a computer program that acted in the role of a Rogerian psychotherapist. ELIZA was a relatively simple program; it was not able to "understand" the text input, but it was able to transform the input into generic questions to keep the conversation going.

The impact of a computer adopting this human role surprised many, including Weizenbaum. Even though people knew intellectually that ELIZA was software, they sometimes treated the program as though it were a human therapist who could actually help them. [1] Fast forward to the 2020s, and we have the rapid popularization of LLMs, indicating people's interest and familiarity in engaging in conversations with AI products. Despite their generative potential, one of my first interests with LLMs was creating prompts to guide people through inquiry processes that could support personal development and conflict resolution.  

One goal for an augmented ELIZA program is thus a system which already has access to a store of information about some aspect of the real world and which, by means of conversational interaction with people, can reveal both what it knows, i.e., behave as an information retrieval system, and where its knowledge ends and needs to be augmented. Hopefully the augmentation of its knowledge will also be a direct consequence of its conversational experience. It is precisely the prospect that such a program will converse with many people and learn something from each of them which leads to the hope that it will prove an interesting and even useful conversational partner."  

Joseph Weizenbaum in ELIZA, a Computer Program For the Study of Natural Language Communication Between Man And Machine

Asking good questions is a powerful tool, not only to access relevant information but also as a mechanism to engage the conscious mind and bring awareness to the autopilot and intuitive behaviors we get stuck with. We don't need to trust or train artificial intelligence to give us life advice, after all, it has no skin in the game, but we can trust it to ask us questions and trigger critical thinking.      

Capy Vera could be an augmented ELIZA, triggering the journaling and self-reflection behaviors needed to understand how we want to participate in society. While it would also provide a sense of support and connection by showing interest in your life and praising your achievements, the strategy of using a cute mascot to suggest tasks to you is a small persuasive tactic to increase your willingness to follow through with them.    

Three dimensions of information

So far, we have been discussing computer-human persuasion, which represents persuasive techniques written as software code and executed by the computer to persuade people. However, interactive information technology also allows computer-mediated persuasion, which refers to ways people persuade others through computers, e.g., discussion forums, messages, blogs, or social network systems. [3]

This is a proposal for a coordination tool, especially one that fosters global collective intelligence and allows the flow of information in a groundbreaking manner. The display of data in three dimensions (self, human connections, and planetary impact) and its organization into different sections within the app can help us navigate the digital landscape. It also supports the big data integration needed for an effective RS and creates possibilities for both computer-human and computer-mediated persuasion.

The Self dimension is where users organize their private information. Following privacy protocols, it is possible to feed the personalized task recommendation engine without violating people's confidentiality rights. As mentioned above, this data is crucial for the functioning of the entire system and the fulfillment of its mission to support massive behavioral change toward the well-being of the whole.

Individuals place equal or greater value on their own interests than on the public interest [4]. Instead of denying it, we must accept the fact and figure out ways for these interests to converge. That is the sole purpose of our recommendation system, and to achieve this confluence, it's helpful to consider the users from a holistic perspective.  

Besides working on use analysis, which focuses on what information is relevant to the user in a given situation, the context analysis is what matters to us the most. It means analyzing a user’s interests, needs, goals, motivations, abilities, pre-existing attitudes, commitment, consistency, compromises, life styles, persistence of change, cultural factors, deep-seated attitudes, social anchors, and perhaps even their whole personality. [3]

One of the most essential facets of analyzing the user context is understanding the user’s goals, including current progress toward achieving them, and potentially past performances. The goal-setting theory acknowledges the importance of conscious goals and self-efficacy, focusing on the core properties of an effective goal and the motivation required for it to work. Overall, persuasive systems should encourage users to set goals and to discover ways for achieving them in a systematic and effective way. [3]

Knowing how to design for behavior change gives you power. [...] all this boosts your ability to change people’s lives. That’s a big deal. I strongly believe the best approach is to help people do what they already want to do. In other words, as a behavior designer, you are not manipulating people or transforming them into someone else. You are helping people become a better version of themselves.

BJ Fogg in the foreword of Designing for Behavior Change [6]

To fulfill this vision of a self-persuasion tool that helps users accomplish their goals, some basic organizational features must be included in this first section, such as a list of working goals, tasks, reviews, and a calendar to help with time management. Ideally, we should work towards the migration and integration of the data users habitually upload on other applications.

By providing a tool that suggests tasks, tracks performance, and transparently informs the results of your use and context analyses, we are following the self-monitoring principle, which states that applying computing technology to eliminate the tedium of tracking performance or status helps people to achieve predetermined goals or outcomes.  

But we are not supposed to figure ourselves out alone. We need peer feedback and community support. As a matter of fact, people can generally achieve a greater degree of attitude and behavior change working together than working alone. That’s the power of social influence. [1] Which leads us to the second dimension.  

The human connections dimension, as the name suggests, relates to our social interactions, collective ideas, and self-organizing communication. It's the section to share our thoughts, intentions, and have discussions. A place to turn our goals and task ideas public to gather peer input and collaboration. It's the instance that allows us to map our support network, to follow friends, and to organize in groups. It's the realm where computer-mediated persuasion starts to prevail and where things get really interesting.      

Don't get me wrong, a self-persuasion tool is great, but alone it only works for people who already have an idea of their full potential, and its self-centered focus cannot help us to foster mutual aid and communal participation. But if we cross individuals' self-set goals, locations, and even their secretive intents and aspirations, we can make connections and local collaboration suggestions that would be otherwise imperceptible.    

Therefore, the key feature of this section is a newsfeed of task offers and requests being uploaded by your neighbors and by people with similar interests and complementary skills. The personalized task recommendation engine is designed to curate this content to help individuals take action in their lives and maximize their chances of making new worthwhile connections. It allows us to know our peers for what they are interested in and willing to do, especially when their profiles match our own self-set goals, and can inspire us to act.    

Let's picture a scenario, Mary and John are neighbors but they have never crossed paths. In the private section, John set the goal of 'losing weight'. He never talked about this with anybody because he thinks it's only a matter of dieting and exercising, and he'll eventually do it. So he refuses Capy Vera's suggestion to share it with peers as a request, even though he's not making much progress, thus this information remains private. A few blocks away, Mary has 'running' as a personal skill. She shares with Capy Vera the data she tracks with her fitness wearable, and the task of running gets automatically updated every time she exercises. So Capy Vera, that intimately knows both of them, realizes the opportunity to be persuasive by applying the social facilitation principle, which suggests that people perform better (more, longer, harder) when other people are present, participating, or observing [1]. So it asks Mary if she is willing to put her passion at the service of others and suggests she share the running task as an offer to her peers. Mary never thought of starting a running club, but she accepts the suggestion just to see if anyone is interested in being her running companion. Surely, Capy Vera then makes sure John sees this offer and that he also knows that the itinerary is closer to his home, that it's happening during a time he is available, and that he and Mary have many other interests in common. With that extra bit of information, Capy Vera is also being persuasive by following the information quality principle, which states that computing technology that delivers current, relevant, and well-coordinated information has greater potential to create attitude or behavior change [1]. Besides all Capy Vera's efforts, John ends up ignoring Mary's offer. A person would probably give up on John, but software designs are persistent. Capy Vera is chilling and thinking: "Maybe he doesn't like running, what else can I try? Oh, a person he follows just shared an interest in cooking healthy. Nice!"  

Hopefully, this tale is enough to help you visualize the potential of a RS designed to facilitate mutual aid and powered by data-centered persuasion and social networks. Networked computing products can be more persuasive than those that aren’t connected because they can provide better information, can leverage social influence strategies, and can tap into group-level intrinsic motivators [1]. Also, the interactive nature of social media could be harnessed to promote diverse democratic dialogue and foster collective intelligence [2].

For that, in addition to the task offers and requests, this newsfeed can also be used to crowdsource information by allowing people to share discussions, polls, and other interactive systems that collect and display informative metrics about the behaviour and reactions of others. For instance, whenever there is a conflict or violation of a group's goal and task agreements, the parties involved can be asked to detail their perspectives by filling out a form with pre-defined questions inspired by non-violent communication, which aims to check their biases, assumptions, and unmet needs against the facts. Each statement can then be shared for public or in-group scrutiny. As uninvolved parties share their opinions through comments, reaction buttons, and voting systems, people are allowed to tap into collective intelligence to reach a synthesis or resolution. In worst cases, people can be sanctioned, meaning they won't have access to any information shared by the person or group applying the sanction. For having a pre-defined process to guide conflict resolution, we are applying the tunneling principle, which states that using computing technology to guide users through a step-by-step process or experience provides opportunities to persuade along the way [1].

In the introduction, it was mentioned that this is a proposal for an information system centered on tag-filled tasks, which are the fundamental blocks of information we compile and organize. I have shared how it works from an individual viewpoint. Now, to understand how the group dynamics play out feature-wise, let's get back to Mary's tale. The 'running' task she offered was shown to her friends and possible matches. Once a task shows up in your newsfeed, you have 2 options to interact, request and/or collaborate. If you click 'request' it means you support that action and you want it to happen so you can benefit directly or indirectly from it (appropriation). If you click 'collab', it means you are willing to work towards it and take leadership roles (provision). To Mary's surprise, she had 10 people requesting the 'running' and 2 people willing to 'collab', so 10 were willing to show up and 2 were interested in helping Mary with the organizing decisions and activities. Once the first people interacted with the task, a group whose primary goal was 'running' was automatically formed. Depending on the chosen interaction, people gained access to group features, like their own chat group, voting features, the ability to create subtasks, a conflict resolution mechanism, editing, etc. So they started self-organizing, and when the first task was complete, they unlocked the third and last dimension.  

The planetary impact dimension refers to the concrete and material things we create and manage. It's the section in which we visualize our completed tasks and have tools to measure their impact, such as reviews, images, videos, testimonials, documents, environmental metrics, or any other relevant data. It's also a place to track and manage our resources.  

The two main features here are the resource inventory and the newsfeed of completed tasks. For a task to appear in this dimension's feed, someone inside the task group must upload a proof of completion, and different types of file attachments should be valid for that. Depending on group settings, the report of their actions can be public or private, and once shared, users can interact with it by providing peer feedback through impressions, learnings, suggestions, etc. Individuals can also be incentivized to share with their network the proof of their personal achievements to gather peer feedback and support (jumping from first to third dimension). Such a display of accomplishments helps us follow the recognition principle, which implies that by offering public recognition (individual or group), computing technology can increase the likelihood that a person or group will adopt a target behavior [1].

Besides exposing the actions of friends and groups you're involved with for your recognition, the personalized recommendation engine can also take into consideration other persuasive principles that leverage social influence to curate this newsfeed. Like the social comparison principle, which says people will have greater motivation to perform a target behavior if they are given information, via computing technology, about how their performance compares with the performance of others, especially others who are similar to themselves. And the social learning principle, which states that a person will be more motivated to perform a target behavior if they can use computing technology to observe
others performing the behavior and being rewarded for it. [1]  Guess who saw the photo of the neighborhood running club Mary started, and got intrigued when he realized they organized a snack table at the end?

Regarding the resource inventory, it's a feature to track and monitor physical resources. It allows users to upload a resource with its proof of existence in order to get it added to the tag system, with its unique tag. This allows individuals and groups to self-organize around the resource, creating tasks to manage it, tracking its location, status, and rights of usage, or adding it to the list of needed resources required for a task. By default, we can have land and its official division into streets, blocks, cities, states, and other natural resources already set into the tag system and resource inventory to support their collective management.  

The combination of these three-dimensional types of information helps us realize the interconnectiveness of our individual and collective thoughts and actions. Individual actions influence the whole, and collective actions influence individuals. The division presented here serves only for organizational purposes; the real intention is to integrate all the information so we can achieve greater collective intelligence. However, we cannot fulfill this intention if we cannot trust the information being presented or if we are not held accountable for our actions. Therefore, one last feature setting is required.

Safety measures

As a global collective intelligence tool, the purpose of this prototype is to support our decision-making by providing an accurate view of the world and the things happening around us. It's a bold endeavor, especially if we consider the rise of fake news, deep-fake content, and bots that are currently flooding our social media. But even before the internet, we have been constantly fed false information by the media and governments (vide documentary Hypernormalization). And when we act based on false information, the results might at least be suboptimal. It's like walking without seeing the real ground we're stepping on; at any time you might stumble or fall into the abyss itself.

Now, more than ever, it's worth trying to ground ourselves on what is real and avoid such a nebulous fate. It helps to remember that, as there will certainly be people lying and deceiving, there will also certainly be people fact-checking and truth-seeking. To support the latter more than the former, some design features can be implemented.

Biometric authentication - to avoid the presence of bots and fake profiles, I suggest access to the prototype be granted through biometric authentication, which uses an individual’s biological traits to authenticate them. Most smartphones now have sensors to collect biometric data, such as fingerprints, facial scans, and voice patterns. This can help implement biometric authentication without much friction. And in doing so, we can have one profile per person, which means that when we're having a discussion or when we submit a topic for voting, we can trust that it's actually a human being we are engaging with, and we don't have bots faking opinions. Moreover, one profile per person guarantees that people are not allowed to hide behind fake accounts and escape the consequences of their actions. However, it's important to notice that we are dealing with very sensitive information, which requires extra security measures to protect users' data, like strong encryption and proper implementation.  

Anonymous mode - sometimes hiding is important and comfortable. It can protect whistleblowers, political activists, and introverted sky people. Anonymity is also important when people are experimenting with new attitudes and behaviors. You may have sensed this phenomenon in anonymous chat rooms: shy people can try being bold, those with conservative values can test liberal waters, and those who normally guard their privacy can open up and speak their minds. For better and for worse, anonymity helps overcome social forces that lock people into ruts and routines. At times, anonymity makes it easier for people to change. [1] Therefore, if we implement one profile per person, it's also crucial to implement an anonymous mode that allows their expression without direct connection to their profile. Nevertheless, the anonymous mode shouldn't be used as a means for impunity; the reports of violations, flags, and sanctions must still be tied to the perpetrator's profile and metrics.  

Peer review - when someone reports an action or marks a task as complete, how do we know it is true? How do we know the proof submitted is not fake or AI-generated? How do we know someone will follow through with their commitment to collaborate? Well, we can ask third parties to confirm and review each action, allow people to flag fake content, and measure people's reliability. These metrics, reviews, and flagging must be visible to people accessing the information, so they can make their own judgment about its trustworthiness and credibility. For instance, if you're collaborating with a person and you know they have 20% reliability (according to the people they have worked with, from 10 tasks they click collab, only 2 they follow through), you know what to (not) expect from them. The same goes for fake content; if a task receives one flag from an anonymous source, it could be an ill-intended or misinformed person, but if it gets 100 flags and angry comments, it's clear that something is going on. These peer review mechanisms are great alternatives to track people's reputation and crowdsource content moderation.

Invite-only onboarding- considering we are designing a cyberspace that aims to collect, store, and share truthful individual and collective information, the way people are invited is of great significance. By applying an invite-only onboarding ritual, we can track the relationships between profiles and map how the network is growing. People would probably only invite those they know and trust, which helps create the environment where we know that everyone is a friend of a friend.      

Final requirements

A system’s persuasiveness is mostly about system qualities. There are a multitude of aspects that need to be recognized when designing persuasive systems, including responsiveness, error-freeness, ease of access, ease of use, convenience, information quality, positive user experience, attractiveness, user loyalty, and simplicity, to name a few. [3] Quite understandably, if a system is useless or difficult to use, or it is not well-mapped with a user’s first and foremost interests and needs, it is unlikely that it could be very persuasive. [7]

This is not an easy task: building a data-centered persuasive information system that is safe and easy to use. But it's not an impossible task either. And the advances of AI and vibe-coding can simplify its development. This prototype idea has lived in my imagination for the last 5 years. It started as a mind map, and I've been refining and trying to build it ever since. I realize these are still assumptions, which is why I've decided to approach it as a behavioral science research project. If we follow this design, can it fulfill its purpose and support regenerative behavior change?

Phase 2: Community Activation & Storytelling

Before I explain the context in which I plan to test the prototype, let me share a bit about myself and introduce you to my friend Carol. Back in 2017, I was thinking like Tupac, and even though I was unemployed, I was motivated to work as a volunteer. I grew up listening to my mother share her life stories about child abuse and how she was forced to work as a house cleaner at 10 years old (a residue of Brazilian enslavement culture), and as I was a young single mom, working with kids felt like a special calling for me.    

So I found Educar+ on a website for volunteer work and applied for a storytelling position. That's when I met Carol. She had mobilized friends, family, and neighbors, collected children's books, donations for snacks, volunteers to cook and read to her neighborhood kids, and a space at the back of her church. It was a simple one-day project, but I was impressed with Carol's passion and realization capacity. She was motivated to do it because someone did the same for her when she was a teenager. When you live in a favela here in Rio, you don't have many educational opportunities. Most kids don't know how to read and are often co-opted by drug dealers. This was enough incentive for us to keep repeating the event every month.        

So we became friends, and I remained a volunteer until we were able to legally register Educar+ as an NGO, get our headquarters, and the financial ability to pay a couple of teachers. Today, Carol coordinates an NGO with 17 employees, 3 volunteers, 7 permanent after-school projects, and 500 children, teenagers, and mothers assisted per year.    

I'm telling this story for a few reasons. Firstly, I believe vulnerability helps to build trust, and it is important to share where this prototype idea (and its biases) comes from. Secondly, Carol and the many volunteers who helped Educar+ throughout these years, including myself, are examples of people willing to work based on intrinsic motivation and motives beyond financial compensation. People like us exist everywhere, and a person with passion and plans can persuade us into action. Lastly, I want to hire Carol as a research assistant to help select research participants within the Educar+ network and implement the prototype in her neighborhood.  

Educar+ has always been a workplace where I felt safe to experiment. When I was studying screenwriting, I taught an audiovisual course there. When I wrote a new fable character, I would present it to the kids. Carol always supported me. Besides, testing this prototype in a favela, a marginalized territory that suffers moral stigma, government neglect, financial limitations, and the presence of drug trafficking parallel power, seems like an interesting political move.    

Despite Carol's superpower of mobilization, I'm not trusting her to single-handedly motivate her whole community to use the prototype. I have another strategy: to spare part of the soon-to-be-crowdfunded budget to provide resources so they can engage it in whatever tasks they want. Meaning, we'll tell research participants they have a certain amount of money worth of resources that they can use to nurture life and improve their community; and that they must use the prototype to self-organize around that and decide which tasks and resources they want to use and accomplish. The most requested tasks in the prototype get the needed resources, which are then uploaded to the resource inventory.

Will the number of users surpass the number of research participants? What kind of tasks will they choose to do? What kind of resources will be requested? Will they engage in tasks beyond those that will be resource-funded? The answers to these questions, along with those from the research questionnaire, will help us assess the effectiveness of our prototype in persuading individuals to engage in mutual aid and communal participation. And all these answers are going to be documented and shared in a publication.

Phase 3: Research Publication  

The features to be developed in the prototype will depend on the success of the crowdfunding campaigns. To build and test the personalized task recommendation, I estimate a $100,000 budget and a 12-month timeframe as simplified below.  

Phase 1: Development of Personalized Task Recommendation Engine (6 months)

1. Data Collection and Preprocessing: tag system creation and regenerative tasks analysis.
2. Machine Learning Model & Software Development: content-based filtering with tags, UI implementation, back-end and software development.

Phase 2: Implementation and Community Activation (5 months)

1.    Recruitment: ~100 research participants from the NGO’s circle of active students, volunteers, parents, and neighbors.
2.   Baseline Data Collection: gather information on participants' current levels of prosocial behavior and community involvement through questionnaires and interviews.
3.  Implementation and Resource Allocation: participants are instructed to use the prototype and receive resource allocation based on their use and collective decision-making.
4.   Data Tracking and Midterm Assessments: Monitor participants' task completion, engagement levels, and conduct midterm surveys to assess changes in prosocial behavior and improve the tag and recommendation systems.

Phase 3: Research Publication (1 month)

1. Final Evaluation and Essay: Gather participants' final levels of prosocial behavior and community involvement through a new set of surveys and interviews to measure the overall impact of the recommendation engine. Write an essay with the research results.
2. Open-source: the prototype's code and other relevant outcomes.

Project Timeline: 12 months

Budget Estimate (with 5 team members): $100,000

• Data collection and evaluation (researcher + assistant): $20,000
• Model Development (AI/ML engineer): $15,000
• UI Implementation (UX/UI designer): $10,000
• Software Development (back-end developer): $15,000
• Community activation (resource allocation): $30,000
• Miscellaneous (equipment, software, hosting, etc.): $10,000

It would be amazing to build the prototype with all the features described in this article and test its full persuasive tactic, but I will be happy if I can accomplish it in any capacity at all. The intention is to build a fully transparent project, and publicize all its completed tasks and outcomes as if we were displaying our work in the 'planetary impact dimension' already. Regardless of the outcome, this is supposed to be a gift for the world and humanity, to help us tune into a different mode of organization (one that is not extractive and competitive).

By open-sourcing all the work and making the tag system, the code, the designs, the budget, the questionnaire's questions and results, and other subproducts available to the public, we can guarantee our effort won't be in vain. Even if the prototype fails to engage people in regenerative behavior, at least sharing our learnings will make it worthwhile.    

Hopefully, after reading this very detailed description of the prototype and research plans, this product is now living in your imagination too and has made you also hope for its materialization. Which might help with the next step to build it: crowdfunding (seems the right way to fund an open-source research project that does not seek revenue).    

If you support my work so far, please consider spreading the word and showing some love by buying me an açai ;)

See you soon during crowdfunding \0/

References:
[1] Fogg, B. J. Persuasive Technology: Using Computers to Change What We Think and Do. Morgan Kaufmann Publishers (2003).

[2] Lorenz-Spreen, P., Lewandowsky, S., Sunstein, C.R. et al. How behavioural sciences can promote truth, autonomy and democratic discourse online. Nature Human Behavior, Volume 4, 1102–1109 (2020).

[3] Oinas-Kukkonen, H., Harjumaa, M. Persuasive Systems Design: Key Issues, Process Model, and System Features. Communications of the Association for Information Systems (2009).

[4] Youngsoo, S., Jinwoo, K. Data-centered persuasion: Nudging user's prosocial behavior and designing social innovation. Computers in Human Behavior, Volume 80, 168-178 (2018).

[5] Bonicalzi, S., De Caro, M. & Giovanola, B. Artificial Intelligence and Autonomy: On the Ethical Dimension of Recommender Systems. Topoi 42, 819–832 (2023).

[6] Wendel, Stephen. Designing for Behavior Change: Applying Psychology and Behavioral Economics. O’Reilly Media (2013)

[7] Oinas-Kukkonen, Harri. A foundation for the study of behavior change support systems. Personal and Ubiquitous Computing (2013).

Find me
Microblogging on Mastodon and Bluesky
@flaviagoma on Telegram
@membranapermeavel on Instagram