You didn’t think we’d left did you? Tank’s been been beefing up AlphaFarm’s backend and soon we’ll have a have a new release. More on that below.
So are IQs a useful metric? Yes, but they’re imperfect and incomplete. IQs represent attribute strength across the genome matrix but may over- or underestimate performance when it comes to a specific activity.
Take take a look at the picture below.
This is Alonzo Clemens, and he sculpted this bull with his bare hands. Alonzo is a world famous artist despite suffering a traumatic brain injury when he was a child leaving him with an estimated IQ of 40. Yet, it’s clear that his ability to mentally represent and rotate objects in the three dimensional space is off the charts — he’s a visuospatial genius. Quite simply, Alonzo is a savant.
If you were to estimate the ASM version of Alonzo’s brain’s visuospatial ability based solely on IQ you’d be wildly off. This is because genome matrices are parceled and each parcel represents a different attribute (this is a simplification but useful here). Because attribute strength varies across the genome matrix, every brain will have strong and weak parcels. So, if most of a genome matrix’s parcels have low values, those low value can drown out a very weighty parcel when calculating IQ.
At AlphaFarm we’re working to provide you with insights beyond brain IQ. We want to help you identify the right brain for the right application.
Borrowing definitions from psychometric literature we’ve analyzed ASM genome matrices and discovered savant genesis brains. Moreover, we’ve identified some exceptionally gifted brains, some of whom have an average or low IQ. But before we get to how we define “savant” and “gifted” we’ll need to introduce the concept of parcel weight.
As you now know, the genome matrix is a heatmap of attribute strengths and weaknesses. Each game decides how to map in-game attributes to regions of the genome matrix. We call these regions parcels. Parcellation of the heatmap was diagramed in the ASM whitepaper and for this analysis analysis we used the same layout — a 4×4 grid of 16 parcels. However, we can adapt to any parcellation design.
So, just as genome weight is the sum of a genome matrix’s attribute levels, a parcel weight is the sum of a parcel’s attribute levels.
Having broken up the genome matrix into different parcels, we used standard psychometric definitions to define which brains are gifted and which are savants on a per parcel basis.
Giftedness is defined as having a parcel weight that is 2 standard deviations above the population average (top 2.5% of brains for a given parcel). Savantism is defined as having a parcel weight that’s 2 standard deviations above that brain’s average parcel weight AND 1 standard deviation above the population average.
When we plot the IQ distribution for gifted and savant brains something interesting emerges: there are high IQ brains that are neither gifted nor savants, and multiple gifted brains with low IQs.
Of the first ~6,500 or so genesis ASM brains, about 4,000 are gifted in at least one parcel. Once all brains have been minted, we’d expect about 250 to be gifted for a given parcel. This means that while many brains are gifted, few are gifted for any given parcel and few brains are multiply gifted. Indeed, ~500 brains are doubly gifted, 65 triply gifted, and only 7 are quadruply gifted (gifted in 4 or more domains). There’s even a quadruply gifted brain with a completely average IQ, about 100.
The savants are even more astonishing. Savant IQs are low and typically range between 50 and 100. Brains that are both gifted and savant have an even lower IQ range (~50 – 75).
In the next week or so we’ll release parcel metrics.
Below shows the 16 parcel genome map by parcel rank and a bar chart of parcel weight to the right. In the new release, these ranks will be displayed as percentile rank, which is a easier to interpret. Percentile rank transforms parcel rank to a percentage (e.g. 75th percentile means that that parcel’s weight is higher than 75% of all other brains).
In time, we’ll also help you find brains with complementary genome matrices so that you can build a team with coordinated starting attributes. Our brain browser will also work for gen2 brains (unless ASM flips the script on us).
In the next post we’ll delve into what makes a good game and how ASM brains appear to have been designed to maximize enjoyable gameplay.
We don’t know about you, but we’re bullish on savants.