Peer Networks and AI in the Building Retrofit Industry

Introduction

In a previous post I looked at evidence-based policymaking in a time of transformational change. I argued that peer networks supported by AI can help steer the global market transformation (that's already happenning whether we like it or not) towards social and environmental health and well-being. In this post, I'll explore this idea in more detail, using the building retrofit industry as a model.

1. The Imperative of Building Retrofitting

Buildings represent about 36% of global energy use and 39% of energy-related carbon dioxide emissions annually¹. Retrofitting provides an opportunity to upgrade buildings to meet current energy efficiency standards, effectively reducing energy consumption and emissions while also making them safer, healthier, and better prepared for the effects of climate change. Governments all over the world have set energy and emissions performance targets for their own buildings and have begun imposing those targets on the private sector, first with subsidies, and soon, with regulation.²

2. Challenges in Retrofitting Ecosystems

The building industry is resistant to change. And it passes that resistance to its child, the nascent retrofitting industry. But government is forcing an industry transformation,^{3 4} which, predictably, is causing the industry to respond in unexpected and, sometimes, counter-productive ways. ⁵

Let's look at the challenges the industry faces across three dimensions: technology, workforce, and policy.

2.1 Technology Challenges

Advances in technologies hold immense potential, but multiple barriers hinder their development and market penetration.

1. R&D Investment Barrier: A reinforcing loop where high initial R&D costs deter investment, leading to fewer innovative solutions, which in turn reinforces the high barrier for entry, especially for smaller players.⁶

2. Complexity and Delay: The technical intricacy of retrofitting introduces delays into the system, slowing the feedback loop between innovation and market introduction.

3. Transition Bottlenecks: The challenge of scaling up from prototype to commercial production is a bottleneck in the system, slowing the flow of new products into the market and impacting the feedback loop of innovation adoption⁷.

4. Boundary Issues - IP Landscape: Navigating the global intellectual property system presents boundary challenges, with innovators needing to define and protect their innovations amidst varying international rules and norms.

5. Existing Technologies as Dominant Attractors: Established technologies pull market attention and resources towards them and away from newer retrofitting solutions.

6. Interdependencies with Ancillary Systems: The success of certain retrofitting solutions is interdependent on other systems (like AI or sensors), creating a more complex web of factors determining the solution's overall viability.

7. Long Feedback Loop for ROI: The extended payback periods introduce a long feedback loop for return on investment, making it challenging to attract initial interest from both developers and end-users.

2.2 Workforce Challenges

While technology advancements are central to the retrofitting industry’s progression, an equally pressing issue is ensuring the workforce is equipped with the skills necessary to implement these innovations. The training challenges in the retrofitting ecosystem are multifaceted:

1. Continuous Learning: The rapid evolution of technology in the retrofitting industry creates a reinforcing feedback loop, where as technology advances, the need for updated training increases, putting a heavier burden on workers in retrofitting than on those in the traditional building sector.

2. Absence of Standards: The lack of standardized training programs can be a system weakness. Without a universally accepted protocol, the flow of skills and knowledge is disrupted across the ecosystem.

3. Resource Limitation: Similarly, the limited availability of comprehensive training materials is a constraint in the flow of knowledge, impeding effective workforce development.

4. Learning Mode Discrepancy: The system displays a misalignment between learning modes. Retrofitting's hands-on nature calls for practical, on-site training, challenging the traditional classroom-based model.

5. Complexity of Skill Integration: Crafting a single training program that balances the diverse skill sets required in retrofitting introduces complexity to the system, requiring multi-dimensional learning approaches.

6. Cost as a System Barrier: The financial implications of high-quality training act as a barrier, potentially slowing the flow of knowledge and skills to parts of the system, especially smaller nodes (firms or contractors).

7. Geographical System Imbalances: The concentration of training resources in specific areas creates imbalances within the system, leading to disparities in knowledge and skill acquisition across different regions.

8. Cross-system Cultural and Linguistic Differences: Cultural and linguistic disparities can be seen as friction points in the system, making the transfer and assimilation of knowledge less efficient in certain regions.

9. Lack of System Validators: Without recognized accreditation bodies, the system lacks important validation points, making it challenging to ensure uniform quality and competency throughout.

10. Resistance to Change: As I described above, a balancing feedback loop exists where seasoned professionals resist new methodologies, slowing the adoption of new techniques, which in turn reinforces their reliance on established methods.

2.3 Policy Challenges

Building retrofitting, being at the crossroads of construction, environmental sustainability, and technological innovation, is deeply impacted by policy decisions. However, crafting policy for such an emerging and multifaceted industry presents several challenges:

1. Absence of Systemic Historical References: The novelty of the retrofitting field implies a lack of established pathways or historical feedback loops that can guide the formation of policy frameworks.

2. Balancing Loops - Regulation vs. Innovation: The system needs to strike a balance between two feedback loops: one where increased regulation enhances safety and quality, and another where excessive regulation dampens the flow of innovation.

3. Inter-system Coordination Complexity: Retrofitting, as a node, intersects with various other systems (energy, housing, affordability, environment). Synchronizing these systems' objectives and actions adds layers of complexity to policy-making.

4. Local Variation and System Universality: The system experiences tension between the need for policies that cater to local conditions and the push for overarching national standards that span across all subsystems.

5. Delay in Policy Adaptation: There's a time lag in the system where policy development and implementation processes can't keep up with the rapid pace of technological innovation, leading to potential mismatches.

6. Stakeholder Engagement as System Inputs: A healthy ecosystem needs diverse inputs (from various stakeholders). However, ensuring the quality and diversity of these inputs, given the vast array of stakeholders, presents challenges.

7. Resource Allocation Dilemma: The decision-making around financial resource flows, like incentives and subsidies, can be contentious, as it impacts the system's reinforcement for retrofitting adoption.

8. Policy Enforcement: Crafting a policy sets a direction, but a feedback loop in terms of enforcement mechanisms is crucial. This ensures that the system's components adhere to the pathway set by the policy without encountering excessive friction or barriers.

9. Equity and Inclusivity Pathways: The system should be designed in a way that promotes equitable flows of benefits. Policies must ensure a fair distribution of retrofitting benefits across all sub-groups, including marginalized nodes.

10. Iterative Feedback and System Evolution: Given the dynamic nature of the retrofitting ecosystem, there must be mechanisms for continuous feedback and adaptation. This introduces a need for a loop that reviews and revises policies in line with system changes and requirements.

2.4 Market Adoption

1. Market Acceptance: Retrofitting is stuck in a reinforcing loop where market skepticism about new technologies lowers adoption rates, further reinforcing skepticism as there are fewer real-world examples to demonstrate reliability.

3. Communities of Practice (CoP) to the Rescue

A Peer Network is a goal-oriented social network, basically a community of individuals and organizations with similar or complementary professional interests, who connect online to share information, provide mutual support, and share resources and opportunities. A Peer Network operates on the principle of give-and-take, where members offer and receive advice, feedback, referrals, and other forms of support from their fellow members.⁸

3.1 Growth and Stabilization

CoPs can be visualized as adaptive systems within the larger ecosystem. When they are healthy, they can greatly influence the behaviour and evolution of the system as a whole.⁹

CoPs can play a vital role in the retrofitting ecosystem, acting as accelerators, enablers, and resilience builders. They harness the power of collective intelligence (and influence) to drive systemic change.

1. Aligning Goals and Worldviews: CoPs can help align the goals and worldviews of peers, ensuring that the industry is working towards a common vision. This is critical for keeping up a general desire for change; and it can also reduce general friction and increase efficiency across the industry.
2. Knowledge Sharing: Knowledge asymmetry acts as a barrier within the system. CoPs can disseminate best practices, innovative solutions, and experiential knowledge more efficiently. This horizontal flow of information can reduce the unfamiliarity and perceived risks¹⁰ associated with retrofit technologies, making adoption smoother acrosss the system. Also, when professionals share their methodologies and results, it can lead to the development of new, or improved, best practices and metrics.
3. Resource Pooling: By collaborating, groups can combine resources to fund research or create tools that individual organizations might not afford independently.
4. Novel and Rapid Collaborations: System change often involves addressing multiple intersecting challenges with singular interventions (i.e. multisolving). CoPs can facilitate mutually supportive collaborations, connecting organizations and individuals with complementary, goals, skills and resources.
5. Expert Review: A Peer Network naturally hosts a range of experts. These experts can evaluate and critique projects, ensuring they are scientifically sound and practical.
6. Standards and Metrics Setting: A CoP can play a crucial role in developing, testing, and evaluating industry standards and metrics. With the collective endorsement of a network, these standards are also more likely to be accepted and adopted industry-wide.
7. Peer Influence: CoPs can create reinforcing feedback loops that amplify positive narratives and success stories¹¹. When one member adopts and benefits from a specific technology or technique, this positive experience can ripple through the community, increasing the likelihood of adoption by others.^{12 13}.
8. Mitigating Bottlenecks: Service providers connected in CoPs can more quickly identify and resolve bottlenecks. For example, if a retrofitting project requires a specific skill set, a peer network can help a peer identify and connect with the right service provider.
9. Feedback and Continuous Improvement: CoPs facilitate real-time feedback. Insights from one peer can be quickly disseminated to others, enabling rapid iterations, innovations, and improvements across the industry.
10. Education & Training: CoPs can organize training sessions, webinars, and workshops at scale, across every niche in the industry, to educate industry professionals about new practices and how to implement them.
11. Resilience through Diversity: CoPs, especially diverse ones, bring varied perspectives, experiences, and solutions. This diversity is a resilience factor, allowing the system to adapt to changing external conditions and challenges more effectively. In essence, when integrated into the system dynamics of the retrofit industry, CoPs can act as accelerators, enablers, and resilience builders. They harness the power of collective intelligence and influence to drive systemic change and overcome inherent challenges in the retrofit adoption curve.

3.2 Challenges with Existing Communities of Practice

While CoPs can be powerful catalysts for change, they are not without their own challenges:

1. Echo Chambers: CoPs can sometimes create reinforcing feedback loops where the same ideas or perspectives are amplified, leading to an echo chamber. This limits diversity of thought and hinders innovation.

2. Information Overload: With everyone sharing and contributing, the system can become saturated. The presence of too much information can make it hard for members to discern quality content from noise, potentially causing decision paralysis.

3. Diverse Goals and Expectations: Each member might have different objectives for joining the network. This diversity can lead to misaligned incentives, which can disrupt the overall cohesion and effectiveness of the network.

4. Network Resilience and Redundancy: Over-reliance on a few key contributors or nodes can make the network vulnerable. If these key nodes leave or become inactive, the system can suffer a significant drop in value or effectiveness.

5. Emergence of Hierarchies: Even in CoPs designed to be egalitarian, hierarchies or influential nodes can emerge. This can lead to power dynamics where some voices are prioritized over others.

6. Boundary Issues: Defining who belongs to the network and who doesn't can be challenging. Too porous, and the community loses its focus; too exclusive, and it may miss out on valuable perspectives.

7. Balancing Adaptability and Stability: A CoP needs to be adaptable to changing conditions and needs. However, too much change can destabilize the system, making members feel uncertain and less engaged.

8. Trust and Reputation Dynamics: Trust is a vital component of CoP. However, in large and diverse networks, ensuring trustworthiness and maintaining a reliable reputation system becomes complex. Misplaced trust can lead to the spread of misinformation or exploitation of members.

9. Resource Constraints: Just as in any system, resources (like time, attention, or even tangible resources) can become scarce, leading to competition and potential conflicts within the network.

10. Interdependencies with External Systems: CoPs do not exist in isolation. They interact with and are influenced by other systems, such as economic conditions, technological platforms, or social norms. Changes in these external systems can affect the functioning and relevance of the peer network.

11. Cold Start Problem and the Network Effect: In the early stages of a CoP's formation, there's a chicken-and-egg dilemma. The value of the community increases as more members join and actively participate. However, if the initial number of participants is too low, there might not be enough value being exchanged to retain these early adopters. This is a balancing feedback loop where the lack of initial value can prevent the growth necessary to produce that value. Overcoming this inertia requires strategies that either artificially boost the perceived value or mechanisms that rapidly scale the number of active participants to reach a critical mass, at which point the network becomes self-sustaining.

3.4 A Role for AI in a Peer Network

A large language model like Llama or ChatGPT can be leveraged in innovative ways to address the challenges faced by CoP:

1. Feedback Loops and Echo Chambers: AI can curate diverse content from a vast database, introducing new perspectives and breaking echo chambers. It can act as a neutral party, asking probing questions or sharing insights from different viewpoints to foster broader discussions.

2. Information Overload: The model can assist in content summarization, categorization, and prioritization, helping members sift through vast amounts of information efficiently and identify relevant content.

3. Diverse Goals and Expectations: AI can provide an onboarding process for new members, guiding them to set clear objectives and aligning them with the right resources or subgroups within the network based on their goals.

4. Network Resilience and Redundancy: AI can serve as a consistent and ever-present resource, filling in gaps when key contributors are absent and ensuring continuous value addition.

5. Emergence of Hierarchies: AI can function as an unbiased mediator, ensuring every member's voice is heard, and providing consistent information regardless of the user's status in the network.

6. Boundary Issues: The model can help in defining and refining criteria for membership, and periodically surveying the community to ensure alignment with its core mission.

7. Balancing Adaptability and Stability: AI can monitor discussions and feedback, identifying emerging trends and needs. By reporting these insights, the network can make informed adjustments while maintaining its core values.

8. Trust and Reputation Dynamics: AI can be employed to develop a feedback and rating system, helping to build and maintain a transparent reputation mechanism for members. It can also provide evidence-based information to validate claims and ensure accuracy.

9. Resource Constraints: AI can help identify and allocate resources efficiently, suggest alternative solutions when resources are scarce, and even help in fundraising or resource generation activities by crafting compelling narratives or proposals.

10. Interdependencies with External Systems: The model can be used to track and analyze external factors that may impact the peer network, offering insights and suggesting proactive measures to adapt to external changes.

11. Cold Start Problem and the Network Effect: AI can boost initial engagement by offering valuable content, answering queries, and facilitating connections. It can serve as an interim participant, providing interactions until the network gains enough momentum.

4. The EarthNet Platform

EarthNet is a platform that hosts (and integrates) CoPs with the help of AI, fostering connections between organizations and individuals based on specific topic areas or projects. It’s a networking tool and a catalyst for action, particularly for emerging ecosystems like retrofitting.

4.1 Facilitating Collaboration Through EarthNet:

1. Topic-Based Connections: With EarthNet, users aren’t just connecting randomly. They are grouped based on interests, expertise, or project needs. For instance, a nonprofit focused on sustainable urban development can find retrofitting experts and vice versa.

2. Project Creation and Promotion: EarthNet isn’t just about making connections — it’s about fostering actionable outcomes. And AI plays a key role in this: creating, evaluating, and promoting projects. The AI can help write project plans, consolidate and summarize feedback, identify potential collaborators, suggest relevant resources, and even help in fundraising. Users can create, join, or promote collaborative projects. A municipality looking to retrofit older buildings can find technical experts and potential funders through the platform.

3. Knowledge Sharing: One of EarthNet’s strongest suits is its capacity as a knowledge repository. Users can share case studies, research papers, best practices, or even challenges faced, ensuring that the entire community learns and grows together. AI can help in content curation, summarization, and categorization, ensuring that users can find relevant information quickly.

4. Staying Updated: Given its nascent stage, the retrofitting industry sees frequent advancements. EarthNet ensures its users stay abreast of these changes, be it new technologies, policy changes, or training methodologies. AI can help write updates and can find relevant content from a vast database, ensuring that users are informed and up to date.

5. Broadening Horizons: While retrofitting might be a key area of interest, EarthNet’s focus on overarching climate action means users are exposed to interconnected areas, and compelling stories, enriching their understanding and widening collaboration opportunities.

6. Feedback Mechanisms: EarthNet also provides avenues for feedback. Users can garner feedback, refining their approaches based on real-world insights, summarized by AI, be it a new retrofitting technique or a policy advocacy campaign.

7. Global Outreach: Retrofitting challenges can vary across geographies. EarthNet’s global user base ensures that solutions aren’t siloed but are adaptable and informed by international best practices.

6. Conclusion

In the complex world of building retrofits, AI-enabled peer network platforms like EarthNet can play a healthy, if not pivotal role. They democratize access to knowledge, foster a collaborative spirit, and, most importantly, catalyze change at every level.

And as society continues to evolve, peer networking platforms have a vital role to play at the heart of this transformation, accelerating and stabilizing change to a post-carbon, ecologically beneficial, and healthy society.

References

Footnotes

1. International Energy Agency (IEA). (2022). Energy Efficiency 2022. Paris: IEA. https://www.iea.org/reports/energy-efficiency-2022 ↩

2. Heerema,D., Frappé-Sénéclauze, T.P., Tam Wu, K., (2017). Energy Regulations for Existing Buildings. Pembina. https://www.pembina.org/pub/decarbonizing-existing-buildings ↩

3. Grandview Research. (2022). Energy Retrofit Systems Market Size, Share & Trends Analysis Report. Grandview Research. https://canada.constructconnect.com/dcn/news/usa/2023/01/new-york-city-on-brink-of-energy-retrofit-era ↩

4. Proctor, T. (2023). New York City on brink of energy retrofit era. ConstructConnect https://canada.constructconnect.com/dcn/news/usa/2023/01/new-york-city-on-brink-of-energy-retrofit-era ↩

5. Zhang, R., Wang, J. (2022). A Review of 10 years Research on Barriers in the Whole Process of Building Retrofit: Stakeholders’ Perception. In: Guo, H., Fang, D., Lu, W., Peng, Y. (eds) Proceedings of the 26th International Symposium on Advancement of Construction Management and Real Estate. CRIOCM 2021. Lecture Notes in Operations Research. Springer, Singapore. https://doi.org/10.1007/978-981-19-5256-2_97 ↩

6. Brennan, T., Ernst, P., Katz J., Roth, E., (2020). Building an R&D strategy for modern times. McKinsey. https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/building-an-r-and-d-strategy-for-modern-times ↩

7. Hyvönen, J. Koivunen, T., Syri, S., (2023). Possible bottlenecks in clean energy transitions: Overview and modelled effects – Case Finland, Journal of Cleaner Production, Volume 410, https://doi.org/10.1016/j.jclepro.2023.137317 ↩

8. Lave, J. and Wenger, E. (1991) Situated learning: Legitimate peripheral participation. Cambridge University Press [https://www.wenger-trayner.com/books] ↩

9. Moutidis I, Williams HTP. (2021) Community evolution on Stack Overflow. PLOS ONE 16(6): e0253010. https://doi.org/10.1371/journal.pone.0253010 ↩

10. Paek, J-J., Hove, T., (2017). Risk Perceptions and Risk Characteristics. Oxford Research Encyclopedias. https://doi.org/10.1093/acrefore/9780190228613.013.283 ↩

11. Kastens, K., (2016). Reinforcing feedback loops power effective Peer Networks. Earth And Mind. https://serc.carleton.edu/earthandmind/posts/feedback.html ↩

12. Laursen, B. and Veenstra, R. (2021), Toward understanding the functions of peer influence: A summary and synthesis of recent empirical research. J Res Adolesc, 31: 889-907. https://doi.org/10.1111/jora.12606 ↩

13. Wikipedia. Technological Adoption Life Cycle. Wikipedia. https://en.wikipedia.org/wiki/Technology_adoption_life_cycle ↩