The THREDD AI Paradigm

A THREDD is an associated thread of learning for intelligent decisioning and discovery. Synchronizing decision making across multiple interdisciplinary THREDDs can therefore be described as a mechanism for developing an unsupervised learning based, artificially intelligent entity.

The letters in THREDD stand for the actions Train, Harmonize, Respond, Empathize, Discover, and Delineate. Each action is guided by one of 6 foundational principles representing the goals of that action.

Train - The Ascendency Principle
Harmonize - The Connectivity Principle
Respond - The Autonomy Principle
Empathize - The Solidarity Principle
Discover - The Compositionality Principle
Delineate - The Predictability Principle

Each principle contributes to the continuous optimization of the decision-making engine. The ability to achieve all goals must be present in order for an AI to be functionally self-sufficient.

The Ascendency Principle - An intelligent agent must continuously seek to achieve a better state of knowledge by improving its existing models with new information useful in achieving its goals. As the data landscape changes, predictability models can quickly become outdated and once optimal decision making engines can see their accuracy deteriorate considerably over time.
The Connectivity Principle - An intelligent agent must be able to access the necessary information from all relevant sources. A decision cannot be made with partial data. If the agent does not have direct access, but can acquire the knowledge through a source that it can access without others intervention, then this will suffice as there is a clear path available to all data.
The Autonomy Principle - An intelligent agent must be able to perform the tasks needed to realize a result. If dependencies exist that prevent this from occurring, the agent is not self-sufficient. Making a decision is not enough. A consequential relevant action must also be taken as a result.
The Solidarity Principle - An intelligent agent must be able to apply purpose and understand the correlation of value in performing tasks that lead to effective decisioning. Its purpose guides its learning and allows it to self-analyze through feedback of its own resulting decisions in comparison to actual outcome. Its purpose also allows it to optimize its own strategy and replace stages required to make those decisions.
The Compositionality Principle - An intelligent agent must be able to conceptualize new models from the components of existing ones without external feedback. It must have an understanding of data and information association as well as some form of perceived relevancy in order to experiment with new strategies through generalized associative modeling.
The Predictability Principle - An intelligent agent must be able to reproduce the result accurately, hence, the decision itself must be measurable. It must be able to compare results of past models with new ones to determine relative predictability and delineate in which circumstances new models are effective to support successive and iterative improvement of its own process.

Applying the THREDD AI Paradigm to Software Development Process

Since the inception of software engineering teams, organizations have tried to compartmentalize aspects of application development and delivery whether it be quality assurance, dev-ops, infrastructure, architecture, performance or reliability. The list goes on and on. As organizations grow these roles become teams, these teams becomes departments, these departments require leads and managers creating ever growing complexity in implementations of today's established software development life cycles. Furthermore, these distinct teams often solve the same problems, but solution from slightly different perspectives. Eventually, process becomes the wall that separates the 'why' and the 'what' from the 'when' and the 'how', or in business terms, the product strategy and vision, from the development and delivery. The foundational premise of this organizational structure still presides in that compartmentalizing should have established a business process model with distinct roles and responsibilities for the sole purpose of doing one single thing. Creating an efficient decision engine. Much like the holy grail of AI, which some have referred to as Artificial General Intelligence, a cross-disciplinary decision engine seeks to find a thread of connectivity across clusters of feedback systems. This is the same goal as any software development methodology. Receiving feedback and productivity metrics across teams allows the business to do just that. Make an informed decision.

Recollecting Conway's Law, an adage stating that organizations which design systems will inevitably design them to mirror their communication structures. A scary thought as engineering teams grow in size and that once smart endeavor to compartmentalize roles and responsibilities becomes debilitating. However, if we build systems that mirror our own structures, then logically those same methods that we apply in AI can also improve our own human processes as well.

In order to build a strong decision engine we can look to the very algorithms used in artificial intelligence today. They are designed to mimic our own thought processes. Associated threads of machine learning models synchronized on a single goal. In this case, that goal is the execution of the product strategy. Hence, the building of cohesive cross-functional teams that discover, learn, train and evolve as development progresses is a necessity for highly efficient product development.

Learn More

To learn more about how the 6 fundamental principles of the THREDD AI Paradigm are applied to software development process, click below and subscribe in the blog section

Thredd-P-Complete