- questions one decided to answer (which says something on its own)
- answers
- declared answers they accept in their partners
- actual partners they pursue (judged by matches, or dates)
While I expect that mostly similarities attract (so-called associative mating), there are compatible traits (e.g one person loves to listen, one person loves to talk), and there is the level of lack of self-knowledge, or hypocrisy (what we SAY we like, vs what we actually do).
And then e.g. probability that person A likes person B can be expressed as:
sigmoid(actualPrefVecA * personalityVecB)
...and with gradient descent magic, we can turn people into vectors!
Which is all very silly since OKcupid will incorporate dealbreaker questions like “how important is your partner’s religion to you” with ones like “do you like horror movies”, and thus you may end up matching at 90%+ with someone whom you would never consider dating.
It used to be much more informative, especially as the answers are weighted. Of course, a much more informative approach would be to use update weights from data.
Also, when it comes to your example:
- I know people with a deep believer and a lukewarm one (I don't take for granted that it is a more important question than if someone enjoys horror movies),
- for important matters (e.g. whole conservative-religious-right-traditional) it is not a single answer but many, many.
Online dating stands on its own in that it is a very front loaded experience in terms of getting to know another human - the judgement begins much before any interaction occurs. That is not the case in any sort of similar degree when you meet someone say at work, or in a sports club, or even in an online game or internet forum.
And as you point out, that makes the notion of a dealbreaker a fairly nuanced one - e.g. I might tell myself that I would never date someone who doesn't want kids because I do want kids, so why would I waste my time? On a website that shows these preferences, I will likely not engage with profiles who mention that they don't want kids, and ignore their messages.
But of course when the new guy shows up at work and he turns out to be really cute and interesting, I may be interested in him/we may start a romantic relationship before these preferences become apparent.
Which can then go in any number of directions - the relationship might fizzle out really fast. Or it may engender personal change for the partners. Or it might lead to many decades of bitter resentment. Who knows! This could make the rational argument that one should accept to chat with anyone on a dating profile regardless of their displayed preferences, but of course that's not how humans tend to reason.
The horror movies one was found to be the best predictor of a successful relationship one actually :)
> Rudder even pinpoints the two OkCupid profile questions that best predict a couple's longevity if they both answer the same way: Do you like horror movies? And, have you ever traveled to another country alone? [0]
You might enjoy https://psycnet.apa.org/record/1982-01296-001 (taken from the comments from the same post but on reddit). tl;dr: proposition that social relations follow a logarithmic trend
It's fun to see folks rederive these things. Multiple arms of psychology have dove deep into this for many years now, including social psychology (individual biases and relationship behaviors), cognitive (modeling decision making processes), quantitative psychometrics (formal mathematical models of how people represent abstract concepts or traits), personality psychology (emphasis on individual differences and patterns), and of course, clinical psych.
Lots of overlap among these, but rather than start from scratch, a bit of reading in almost any of them, perhaps starting with behavioral economics or social psych, might enhance the "vectors".
Rederived from... what? A comfy armchair? The blog post chose to use vector math as its starting point but I’m not seeing any data that validates the approach in anyway, or anything that justifies it besides the assumption that “ You can take preferences and combine them into a single value/point on a vector” - which you can, but how meaningful is it really to say that my “dog preference” is 0.9? Why not model people with irrational numbers, regular n-gons in 24 dimensions, or a directed acyclic graph instead?
I was recently thinking about this, but on a slightly bigger scale.
There’s a term “off-kilter”, which is easy to explain using vectors like this.
If we take the general vector of society, just sum up all personal vectors and normalise, we form a big vector for society. This is what’s “normal”.
Kilter refers to the concept of how aligned we are with society’s vector. So the concept of being off-kilter is how skewed you are wrt this normal vector essentially.
Of course valid for any of the eigenvectors corresponding to subfields again, and this also goes some way to help form the overton window, which has recently been up here in some posts...
Society's normal vector could be whats off kilter :)
Society feels like multiple interacting vector fields ala magnetic and electric fields. But instead of 2 fields there are probably many...personality, knowledge, energy, needs etc. And you exist as some charged particle (negative or positive?) thrown into the middle of all that dynamic chaos being pushed and pulled in various directions.
theoretically if you add the value-vector of all people together you end up with a long vector that represents the direction of society (even the opposing ones), and you can then normalize this to have the unit vector of society. i.e. what is the "direction of society".
Your skew wr.t. this one (inner product with, or projection on to this) will be some number between -1 and 1 (if we account for opposition I guess)
Basically if both vectors are of unit length, you get the cos(angle between). Completely off kilter would be a score of 0. While "opposed" would be -1
In sociology anyways, there's no such thing as a "normal" vector. People talk a lot about the Overton Window in a 1d sense, either you're left or right. When in reality it's a series of multi-dimentional "value" axes which inform / create certain societal norms.
Surveys of this sort are deeply flawed because despite their best efforts they never capture anything close to the full dimensional of the person taking the survey. Asking complicated questions that have "it depends on the situation" answers but asking the person being surveyed to pigeonhole their answers into one of a few crude categories is inherently and hopelessly flawed.
Honestly, this article is fascinating to me for reasons that might discomfit its author.
Not that ignorance of the last 300 years of philosophy on this topic is something necessarily to be ashamed of, but this essay, and the reaction to it here, is of interest to me because (a) this is not an empirically-validated, or even merely well-argued, approach to reasoning about people, but (b) seems to be thought well of anyway, reading the room in this comments section.
I understand better the class of ontological and epistemological elisions system-thinking folks tend not to 'see', and I am reminded of the (now thankfully bygone) era of analytic philosophy where you couldn't make an argument in a respectable journal without translating your propositions into a symbolic logic (your choice of which).
An excellent kettleball to level up with. It's like geocaching but with unsupported assertions.
I fully admit I know nothing of the actual research in psychology/related fields, and this is just my current understanding built up from when I was a kid (completely anecdotal). I'm primarily interested in people's thoughts in general (both good and bad) as well as where I might be able to improve my understanding with the research that's actually been done.
Hm. What if we said that "these preferences exist and theoretically could be mapped to a vector", but also acknowledge that we're impulsive and won't ever actually be able to rigidly identify our own?
"Preferences" here might be more accurately stated as tendencies with some motive.
It'd be nice if vector based approaches (think MBTI tests being originally created using PCA) would use state of the art dimensionality reduction techniques instead of stuff from the 70s so that way each vector actually matters and explains far more variance in the data.
If you need interpretability just use an autoencoder or simese network instead of PCA. This way the new "personality" vectors are highly meaningful and visualizations are better...
You know throwing a neural network at things doesn't just magically make them better... lot of advantages to linear techniques... not even clear that there's enough structure in a big 5 questionnaire to benefit from a fancy NN... also dubious that autoencoder features are particularly interpretable...
This is what the mypersonality dataset collected on Facebook was. The 5 axes of OCEAN - Openness, Conscientiousness, Extroversion, Agreeableness and Neuroticism for all those who took the test.
I have thought of Personality Vectors similar to this concept but for a completely different purpose.
If we had a Chatbot that would interact with someone hypothetically we could have this kind of distinction between Personality Vectors to mutate its own behavior and to interact with others.
Its sort of the idea I had when I watched the movie Interstellar and TARS is told to decrease his own humor. So if the system would just know who its talking to it would create a preference vector for each person it interacts with.
But you would need some kind of baseline so basically the Personality Vectors would be set to neutral. And each interaction with a Personality vector would be a system that stores a preference vector for each person.
As the system time evolves (which is also in the article) you would basically end up with a set of graphs. There is a notion out there which is called a Dynamic Network analysis [1]. If you also stored the history of that you could then not just be able to rewind the system backwards to a previous checkpoint.
The way you could start learning how to learn a personality vector is a reinforcement learning situation. If you say "bad AI I didn't like that" the system would then look at what it said and mutate the preference vector for that person. Or you could just change the vector manually by a technique that was illustrated above by just saying "reduce humor by 70%".
Then you could have model which would do conditional generation [1] of text based upon interactions with others. It would be basically a reinforcement learning system paired with a generative text model. The reinforcement system would store the Personality and Preference vectors for each person. For each behavior and preference you would need a corpus to bootstrap the system. The reinforcement system would be a retrieval network and it would retrieve a pertained system based on user(s) input and get a generational text network to provide a response. This would be similar to Alpha Go's design. [3]
Interesting article, and good timing - A few days ago I wrote about how we can build decentralized social networks where you can filter your world to see content from those with similar personality vectors to yourself - https://adecentralizedworld.com/2020/06/a-trust-and-moderati...
The idea of preferences as vectors seems like it has some fun consequences.
- HIPSTERS: If you really like something, but don't want others to crowd your interests, you will pretend to be nonchalant about it.
- AUTHENTICITY: A person's interests in a topic can be vetted as "authentic" if they are influenced by others whose average of personality vectors matches the person's.
- SOCIOPATHS: You can make in-roads with a group of interest by transforming your PV; see AUTHENTICITY.
- SKEW: You can influence others more by projecting a _very_ high value for a specific topic. If you love a particular form of music to an absurd degree, you're more likely to convert others.
I don't have any non-anecdotal evidence of any of these phenomena. Rather, these are some reasonable if not amusing deductions from the author's model.
Even more, since for questions there are:
- questions one decided to answer (which says something on its own)
- answers
- declared answers they accept in their partners
- actual partners they pursue (judged by matches, or dates)
While I expect that mostly similarities attract (so-called associative mating), there are compatible traits (e.g one person loves to listen, one person loves to talk), and there is the level of lack of self-knowledge, or hypocrisy (what we SAY we like, vs what we actually do).
And then e.g. probability that person A likes person B can be expressed as:
sigmoid(actualPrefVecA * personalityVecB)
...and with gradient descent magic, we can turn people into vectors!