The Quest for a Skill Spine
Almost every company I’ve been a part of or interviewed seriously for either already has or wanted to create a master skill/standards list — commonly called a skills spine — to be able to used for content creation and learning resource alignment.
Here I’ll share the basic idea if you’re unfamiliar with the concept and also some ideas I have going forward that I’ve been thinking/iterating on the past year or so. As always, I’m always intrinsically thinking about how to express these ideas in the CASE format, because without linking machine readable standards within themselves and to other frameworks I feel stand-alone skills frameworks are essentially useless in todays Cradle to Gray edu-work ecosystem.
What is it
There are certain skills that are commonly agreed upon for learning academic subjects. For example, everyone agrees students need to learn how to count by ones, and be able to show that fractions are equivalent to decimals.
But when it comes to state standards, some states have unique standards that no one else has, or have modified standards to address certain distinctions educators in those states felt important. This is good, but creates issues for publishers and edtech content vendors trying to serve all needs in a cost-effective way.
Examples
From personal experience in PK-12 education of course the Common Core were an attempt to create a common set of standards; most states adopted it at one time and have no either renamed them or states like Texas and Virginia never did and kept their own standards.
ACT published the ACT Holistic Framework in 2017
Open Skills Network is developing a common set of workforce skills with definitions, by domain.
LightCast has published OpenSkills for several years; essentially an ontology of skills with links to wikipedia definitions.
Approaches
I’m using a simple standard from Headstart for this demonstration blog, but it’s things I’m thinking about how to best build a taxonomy for PK-12 or workforce skills that is both flexible, easy to adapt, and can contain connections between skills and topics as they grow and change. A couple of these came from HeadStart blogs, but for others I wanted more granular skills such as including Skip Counting.
Skills with Sub-skills as child objects
First of all, AI can certainly help create ideas for structure and the sub-skills themselves. here is an example of sub-skills ChatGPT created when I prompted it to do so with an example and some refinement. We would call this decomposing a standard into it’s sub-parts.
In the CASE specification this would have the P-Math.1 as a type ‘standard’ and the skills underneath would be child objects, like in a standard tree.
Stand-alone skills with Rubrics
Another approach is to keep the components of the standard not as individual skills, but as part of the overall standard as a rubric-based progression. This too can be expressed in CASE, but I haven’t seen many implementations that show rubrics well. Once I had the skills for P-Math.1, I worked with Claude to create the skill progression table (just to switch it up a bit).
I’m still playing with both ideas — there are advantages of a standard-sub-standard approach when it comes to tagging content, but there are also disadvantages like a huge array of standards — does anyone really want to see over 500 sub-standards for a particular grade level? With a rubric-based approach, it’s essentially creating sub-skills but containing everything automatically in a progression for that standard, which could help adaptive systems and students alike know where they’ve come from and what topics are coming up next.
