"Six Degrees" memory feat - memory method?

7 posts / 0 new
Last post
#1 7 October, 2012 - 13:24
Joined: 6 years 1 month ago

"Six Degrees" memory feat - memory method?

I'd like to memorize some "Six Degrees"-type connections for movie actors. For those not familiar with the "Six Degrees" concept, it involves two actors (usually, but n
ot always, one of them is Kevin Bacon) and the attempt to connect them through their movies, in 6 or fewer movie roles.

For example, if someone challenged you to connect Brad Pitt to Will Smith, you might reply that Brad Pitt was in "The Curious Case of Benjamin Button" with Peter Donald Badalamenti II ( http://www.imdb.com/name/nm0045924/ ), who was in "Bad Boys 2" with Will Smith.

What I have in mind is putting together a list that would initially consist of, say, 10 celebrities, and memorize connections for any pair of these celebrities. The list could grow from there, of course.

As for finding the connections, there are readily-available tools, such as the Oracle of Bacon ( http://oracleofbacon.org/ ) and even Google ( http://www.google.com/search?q=Will+Smith's+Bacon+Number ).

A quick couple of notes:
1) Obviously, you'd never have to memorize a celebrities connection with themselves, just the other celebrities.
2) Once you have a connection between two celebrities memorized, you should be able to traverse the connections either way (Memorizing the connection from Brad Pitt to Will Smith means you also know the connection from Will Smith to Brad Pitt, since it's just the reverse).
3) The above two notes mean that for 10 celebrities, you'd have to memorize 45 connections ((10 * 9)/2). The amount of connections for X number of celebrities would be (X * (X - 1)) / 2.

The question remains, however, what would be the best memory system to memorize this? There's a definite advantage to the use of celebrities in this, in that the faces should already be familiar, but what about the movie names, the pos
sibly-unfamiliar people in the middle (Let's face it, Peter Donald Badalamenti II is hardly a household name), and the connections between all these facts?

Any suggestions for memory techniques for such a feat? | I'd like to memorize some "Six Degrees"-type connections for movie actors. For those not familiar with the "Six Degrees" concept, it involves two actors (usually, but not always,
one of them is Kevin Bacon) and the attempt to connect them through their movies, in 6 or fewer movie roles.

For example, if someone challenged you to connect Brad Pitt to Will Smith, you might reply that Brad Pitt was in "The Curious Case of Benjamin Button" with Peter Donald Badalamenti II ( http://www.imdb.com/name/nm0045924/ ), who was in "Bad Boys 2" with Will Smith.

12 October, 2012 - 21:11
Joined: 2 years 8 months ago

That sounds like a difficult project. Maybe you could build a memory palace with a room for each actor? The Will Smith room could contain every major actor who has been in a movie with him. If you start with the most popular actors, you'd probably be able to answer the most common connections.

Or instead of building a palace, maybe just use the movies as locations?

I'm really bad with actors and movies. If someone mentions a famous actor there's probably a 95% chance that I don't know who they are even if I've seen them in a few movies. If I watch a movie, I can watch the same movie a couple years later and not be able to tell if I've ever seen it before. I don't think I would be good at this. :)

9 July, 2014 - 18:02
Joined: 3 years 4 months ago

I like this problem you've set out :D

This is a hugely difficult graph problem, and I doubt that you will be able to find a way to algorithmically encode this in your mind without going insane. :P Most of the techniques that we use explicitly tend to be sort of linear, however the way that our brains function is very non-linear and interconnected. So it is with that idea that I think there is not a good way to explicitly understand how your brain traverses the graph. However... There is a way to pose the problem that gives it some structure:

You are given a graph.
Given that each node is a movie, and
each edge is an actor (connecting the movies he/she played in),
and the only action you have is traversing edges,
construct a method to get from a random node to the Kevin Bacon node while using only memory palaces.

You could do as Josh suggests, and create a palace per actor. The palace would contain all of the movies he/she's starred in. Each loci would be a movie, and would also contain each (major) actor that starred in it. The loci would then link to other actor's palaces. This would require a lot of scanning and guessing.

Another approach that I think might be promising, is not guaranteed to work, but it significantly reduces the scope of what you have to memorize. You could create a palace just like in the last solution, except you start at Kevin Bacon. Each loci would be a movie, and would contain each actor in the movie. Then, with THOSE actors (with a Bacon number of 1), create branching palaces for each. This way you'll be close to the solution because you'll have covered the inner portion of the graph rather than trying to memorize every movie in existence :)

22 December, 2014 - 04:26
Joined: 2 years 11 months ago

I think this is a very important question. Our brains are largely a neural network so why is it hard to memorize networks? Method of Loci for all its virtues still seems to be limited to memorizing sequences.

My model of this problem is the inverse of jd3johns but still reduces to much the same solution.

- each node is an actor
- each edge is a movie they were both in

In software you represent a graph as a set of nodes and a set of edges. In my model the edges would be the tuple:

(Actor1, Movie1, Actor2)

Memorizing the graph basically comes down memorizing this set of edges. So what is the best way to do this? The first solution which jd3johns uses separate memory palaces to model each side of the transitive relationships:

(Actor1, Movie1) and (Movie1, Actor2)

The advantage of this is if have if you the following fully connected graph:

(A1, M1, A2)
(A1, M1, A3)
(A2, M1, A1)
(A2, M1, A3)
(A3, M1, A1)
(A3, M1, A2)

it can be reduced to a more memorable structure:

(A1, M1)
(A2, M1)
(A3, M1)
(M1, A1)
(M1, A2)
(M1, A3)

Another possibility would be one memory palace per actor where each loci contains both the actor and the movie they were both in. This is the same as the fully connected graph above but requires you to remember two items at each loci. Although the transitive memory palaces seem simpler, my feeling is that this triplet version is easier to traverse, albeit more difficult to build.

A third option I've been experimenting with for networks is to have loci with multiple entry and exit points, each one representing the endpoint of some edge. This maps the network directly onto spatial memory but the problem being that it is difficult to remember more than 3 routes into or out of a locus and you need something like method of loci to remember them so you are back at square one which is just using method of loci to memorize the edges.

For my memory palaces I'm using 3 routes for each locus (back, forward, down) which lets me construct hierarchies and mind maps. Although I'd like to be able to map networks I don't have a practical solution yet. Many networks can be converted to hierarchies so I don't think my quality of life is going to suffer from this failure in cognition...

The fourth possibility is that natural memory IS the best method for remembering networks. I found this when trying to encode melodies for better memorization only to conclude that the best way to encode melodies is using melodic (aural) memory and the best way to encode rhythm is using rhythmic memory (possibly the same thing as muscle memory). The problem with natural memory is of course hooking into it. How do you extract that network of information that is in there?

28 May, 2016 - 11:14
Joined: 4 years 6 months ago

I wonder to what extent it would be possible to reduce some of the brute force aspects of the challenge.

In the most extreme case, a person might set out to memorize every movie ever made and every actor that played in each. Obviously, this would be a ridiculously hard task, but it makes two problems immediately evident: the time and effort to memorize all the information would be unsurmountable, as would the time to identify any paths that would lead from a given actor back towards Kevin Bacon.

Of course, there are other problems, too, like the effort and time required to even compile such a list in the first place, but the problems of storage and traversal (path identification) are probably the most fun to discuss.

The challenge would be to identify ways to reduce the need for brute force in solving this kind of problem. I loved Sam's observation about our brains being neural nets trying to memorize networks. This leads me to wonder how our brains might be leveraged to provide ways to avoid brute force techniques.

I am reminded of ways chess masters avoid using brute force in solving extremely difficult chess problems. For example, one skill young chess players are encouraged to build early in their careers is the ability to quickly and unfailingly recognize a number of archetypical checkmate postions. The idea is to burn certain classic endgame and checkmate scenarios into your brain to the point where they almost jump out at you when you are looking at the board. This often allows a player to suddenly see a potential solution, bypassing a lot of brutal calculations.

28 May, 2016 - 11:12
Joined: 4 years 6 months ago

To continue with my last thought, one way of applying this idea might be to avoid memorizing all possible connections and focus one's efforts on memorizing links that tend to have a higher payoff.

I wonder if focusing on a pareto-like principle might still result in a significant number of wins, while reducing the storage and retrieval/traversal challenges. For example, if you focused on only the a) actors who are in the largest number of movies and b) movies with the largest number of actors from set a, would that be more efficient, resulting in an appropriately large number of wins?

21 June, 2016 - 08:01
Joined: 4 years 9 months ago

I think one method would be to find a common step or two that most actors share. This can be a nexus point that each actor must pass through to 'join' up with another actor. That way, you only have to learn the 2 steps that it takes for the chosen actor to get to the nexus point and from there the few steps that it takes actor B to get to the nexus point but obviously you would say the 2nd half in reverse.
Drop me a mail at [email protected] if you want to chat a little more, I'm in the green place a good bit and we've corresponded before :-)



Download our free ebook! Just click the "Sign up" button below to create an account, and we'll send you a free ebook with tips on how to get started.

Related content: