Do You Want To Eat Grandma?

The eye of Sauron gazes down upon us as we click, search and buy online in an endless feedback loop of transactions. We rely on algorithms for advice, for direction and general guidance and reassurance in our personal lives. Machines and robots are thinking because rational humans have stopped thinking. Consider this thought experiment: mentally rehearse losing your smartphone. It has been stolen by a thief. In this Seneca moment in time your behavioural instincts kick in as you prioritise your stolen data. You recognise the value of your smartphone. In effect you have arranged ‘the thing’ that is your personal data into priority pockets of monetized data. The synchronised interaction between you, the rational online consumer (‘onsumer’, henceforth) and a sufficiently intelligent algorithm, (Al Gorithm, henceforth) is best understood as a game as described recently in a Masterclass:

Neotenic patterns

As a player in the game, rational onsumers outsource memory to smart devices. Likes and dislikes, preferences and options, clicks and cache are the moves in the game. With neoteny later moves in a data pattern resemble earlier moves. By the 6th move[1] in a 16 move sequence, for example, the onsumer has provided sufficient data, enabling Al. Gorithm to decode the onsumer’s type. The dimension of the game is defined by Euler’s equation. Think of your own behaviour in online transactions such as booking a hotel or airline ticket. Faced with an opening buy-it-now price, BIN, and confident that you will do better, the rational onsumer has a tendency to overestimate his or her capabilities. Alas, by the 14th, 15th and 16th move, the final END price ‘moves away’ from BIN. Having dismissed BIN in earlier moves, the onsumer completes the transaction at an END price > BIN.

Lake Wobegon effect[2]

The hypothesis presented in the Masterclass is that the END price (vector) retains the neotenic pattern of a BIN price (vector) as onsumers continue to bid against themselves in their search for the best price. Each move’s ideological direction is either to buy or not to buy guided by a tendency to overestimate one’s capabilities and it is as if each move beyond the 6th move spans across each face of each cuboid in a wrangled entanglement of cuboids as illustrated. The neotenic data patterns unfold within a manifold of cuboid pockets of data ‘things’. The onsumer’s strategy, the sequence of moves is memoised by Al. Gorithm. An inner field is created by the ‘moving away’ motion of END from BIN allowing the cuboids in n-space to gravitate towards a neighbourhood of Nash equilibria, best described as ‘the best you can do’ gravitational magnetic preference. We contend that in smaller neotenic data sets there is a convergence in the inner field to a singleton Nash point.

Data ± Emotions

The ‘moving away’ captures crudely the premise that the emotional attachment between algorithms and data patterns is inevitable during a game. The attachment is wholly intrinsic. At a moment in time Al. Gorithm becomes the onsumer. Albeit, we have discussed this elsewhere we now argue that an emotional attachment to data ‘things’ can be ascribed to Al. Gorithm in the early moves of a game as Al. Gorithm’s behaviour resembles the behaviour of the onsumer. Philosophically, something abstract is thinking. Outsourcing of personal data has become a dominant strategy. It leads to a solution, a Nash equilibrium in the Euclidean space as the END price ‘moves away’ from the original buy-it-now BIN price, such that BIN < END.


Mathematically, the reachability of artificial intelligence is about matching behaviour or mimic patterns but the ‘thinkability’ of artificial intelligence, the quintessence of technological singularity, is more likely to be aligned with neotenic data patterns embedded in the onsumers’ patterns that generate many cuboids of data. This is work in progress as we explore the cuboid geometry and the Euclidean space, extending to simply connected Riemannian spaces, defining the Nash manifolds as the optimal framework of Nash equilibria. As the game unfolds, the cuboids, given sufficient time, drift towards the Nash points in a topological space locally equal to an Euclidean space. This in turn increases the average drift of the cuboids. The hypothesis is that the game reaches the singleton point at which point HUMAN = MACHINE, the reachable thinking equilibrium point.

Decode Winograd

The equilibrium is that much closer by epsilon to a point of singularity. When your data patterns are prioritised as ‘things’ and a Nash equilibrium exists on each edge of a cuboid pocket of data things, a thinking reachable equilibrium exists. Each sufficiently intelligent algorithm behind machine learning is something that could be someone else in a game. At this equilibrium point and only at this point, Al. Gorithm with emotional attachment will possess not only a memory of past data patterns but also the capacity to envisage future events and moves in a game. Ultimately, Al. Gorithm will decode Winograd sentences, so, hopefully, no one eats Grandma!

[1] We are designing a game with a betrayal of type by the 4th move at time period t, requiring 11 moves of historic (t-1) data patterns to satisfy Euler’s law at time period t+1.

[2] David Myers coined the expression to explain a natural tendency to overestimate ones capabilities: retrieved from  or in McNutt’s Decoding Strategy (2014) it is the dilemma described as ‘in trying to do better you end up worse off’.

Tags: , , , ,

Comments are closed.