SimScore API

Wwithout clear prioritization, Big teams become overwhelmed by the sheer volume of ideas, leading to confusion, delays, wasted effort and inaccurate decisions.

SimScore is a prioritization API that connects with current client applications (Ex. Forums, Boards, Surveys) to capture large volumes of ideas. The SimScore Tool calculates a clear prioritization directly from the similarities of the ideas within the dataset. The priorities are displayed to users within the client application.

Big Teams / Communities can focus on what is most important, reducing delays and wasted effort to create and choose better proposals.

The overall effect is more accurate decisions in a timely manner, while reducing delegate fatigue.

Example Integration (SimScore API and Discourse)

A DAO is using Discourse Forum. There are many topics and nested replies. Overwhelming delegates due to the volume and diversity of ideas. To further complicate things direct messages are often needed to gain consent, further overwhelming delegates while also effecting transparency.

A SimScore integration uses existing user replies from the DAOs forum as the dataset. The clustered / priority list along with some advanced featuresare displayed in the DAOs Forum maintain the DAOs user experience.

That’s it, Users can now interact with replies in chronological order (scattered) or in priority order. Leveraging priority order, clients can debate and form consent on the most important issues, saving the time, energy and effort that overwhelms them in the first place.

Key Benefits (Value Driven)

1. Speed
   1. SimScore API calls are processed in seconds.
   2. Immediate Focus: Helps teams quickly zero in on the most relevant or pressing points.
2. Frictionless Adoption
   1. There is no secondary workloadneeded from users or admins. After writing their reply or idea, users do nothing else. Other apps, require up/down voting, drag / drop ranking or pairwise comparison, etc.
   2. There is no change to how users interact with the applications they are currently in using
   3. There is no cost of new foundational software
   4. SimScore uses existing user data, from established applications.
   5. Users do not transition to new software, there is zero organizational disruption.
   6. SimScore Extends existing software’s capabilities--including intelligent prioritization as a feature within the current application
3. Accuracy
   1. Unbiased aggregation is proven to be more accurate that individual experts
   2. No Hallucinations or tweaking.
4. Transparency
   1. No Black Box, As described in the solution section, Unlike LLMs, the pairwise similarity score matrix is totally transparent and easy to audit.
   2. Trust & Verification: Easy for stakeholders to understand how priorities are generated.
5. Low Cost & Customizable
   1. Each idea has a numerical profile which is efficient and pre-computed
   2. Preparing SimScore Prioritizaton to be Token and On chain ready
   3. UX is unchanged, users do not experience a transition, they simply continue using the Organization’s current applications
   4. The SimScore API is live. Connecting to Client applications is available immediately.
   5. Minimal computing cost—unlike LLM.

To streamline decision-making in large teams and communities, helping them navigate and overcome the overwhelm of information overload—seamlessly and without friction of adoption.

Imagine a long Forum Topic or Miro workshop. 100’s of ideas. Scattered chronologically or visually. The impactful ideas are hidden in plain sight, obscured by less relevant content. Big team are overwhelmed leading to confusion, delays, wasted effort and inaccurate decisions.

Within this context, the main problem we target is the feeling of being overwhelmed. SimScore targets 2 aspects of this problem.

1. Identifying the impactful ideas—prioritizing , so team members can discover them without delay or significant effort. This will concentrated debate on the most important issues. As a continuous process, decision making within the big teams will be focussed and iterative leading to faster and more accurate decisions.
2. Teams are already stretched thin. The solution to prioritize the impactful ideas must not add additional delegate workload. Any friction risks rejection due to already heavy cognitive and operational load. So our prioritization tool must strive to make life easier for decision makers and in no way add friction towards adoption (refer to Key Benefits).

In order to solve the prioritization problem, we followed the concept of unbiased aggregation. Unbiased aggregation is one of the key tenets of decision hygiene described in Kehneman’s book—Noise - A flaw in human judgement. For Quantitative (Objective) judgements unbiased aggregation is easy. Simply take the average or vote.

Our goal is to prioritize ideas that are Qualitative (Subjective) Judgements. The unique idea we discovered is the use of a pairwise similarity score matrix to determine the “most similar” ideas and sort them in priority order.

The beauty of the SimScore prioritization solution is that it requires no additional delegate workload. Traditional methods require manual facilitation, up/down voting, or the challenging task of pairwise swiping (which grows exponentially with the number of ideas)

We’ve been experimenting with SimScore within the Arbitrum and Scroll DAO Forums by sharing SimScore-prioritized topic lists. Delegates have responded positively, often referencing these lists to refocus their discussions and debates. This interaction has demonstrated the value of SimScore in streamlining conversations and driving more targeted decision-making.

Details of the SimScore API Solution

Key Terms

Similarity Score - Similarity Score determines the similarity of a pair of ideas.

Pairwise Similarity Matrix - SimScore ranks the replies in “most similar” order,

Clusters - Groupings of ideas that “best fit”

Relationship Graph - A visual representation of the ideas relate to each other.

LLM vs Similarity Score - LLM’s have potential application in the prioritization space. There are currently some drawbacks. Sometimes LLMs are wrong. They are a blackbox. They are often not repeatable.

Using Similarity Score, an NLP tool, offers advantages.

The output is exactly what an author wrote in their reply. An exact paragraph, maintaining the Author's nuance.

Given the same inputs, SimScore will always provide the same output. Repeatable

Being transparent in it’s calculations, SimScore improves accountability. Transparent

Specifications

SimScore Analysis

(presented in Notion)

Clustered Priority list

For each cluster - there is a toggle to show the ideas within the cluster, in priority order.

Please open Demo to view an example of a SimScore Cluster / Priority Report.

Demo - Arbitrum (From MVP Forum) - Demo Report.

Advanced Features

Relationship Graph is a visual representation of ideas and replies as they relate to each other.

1. https://sim-score.vercel.app/session/678fbc9bdafa9331ec0a69b0 Relationship Graph

Pairwise Similarity Score Matrix—Underpins the SimScore API, calculating the similarity of pairwise ideas to determine the” most similar” and sorting in priority order.

1. Pairwise Similarity Score Matrix

Pay for use or SaaS

Want to grow through just grant funding

10,000