Honestly, I’m tired of looking at my timeline.

It’s filled with either people flexing relationship goals that are actually just filtering the best moments (a.k.a. cherry-picking data), or people angsting because they got hurt by the “same type of person again.”

This is a classic issue in software engineering: you keep deploying the same code in a different environment, then expect different results. Einstein said that’s the definition of insanity. I say, it’s the definition of a “lack of algorithm.”

Humans are weird. We use data to buy stocks, statistics to predict the weather, and even Machine Learning to determine traffic routes. But when it comes to choosing a life partner (the most crucial decision that will determine your happiness metrics for the next 50 years) we revert to the Stone Age method: “Follow your heart.”

Bullshit.

“Following your heart” is actually just an accumulation of cognitive bias, pheromones, unresolved childhood trauma (read: technical debt), and social validation. That’s not intuition; that’s spaghetti code running on an operating system called the Mammalian Brain.

As an Applied Mathematics student who chats more with VSCode than with human beings, I feel there needs to be a logic intervention here.

We need to stop seeing love as "magical destiny" and start viewing it as a Stochastic Optimization Problem.

The Legacy Code: "The Ideal Type"

You definitely have a friend (or maybe it’s you) who says: “Ugh, my type is someone tall, fair-skinned, plays guitar, humorous, financially stable, loves cats, pious, but a little bit of a bad boy.”

In data science, this is what we call Unstructured Data. Garbage.

You are piling up qualitative variables without hierarchy, without weights, and without clear operational definitions.

What is “Financially stable”? Minimum wage or having passive income from blue-chip stock dividends?
What is “Humorous”? Cheesy WhatsApp dad jokes or sarcastic US stand-up comedy?

If you input a query like that into a search engine, the result will definitely be zero results. Or worse, the system will force a search for “approximate matches” and you end up with a false positive: A guy who is “sort of a bad boy” but turns out to be just an unemployed guy who’s too lazy to shower.

The problem is, humans are complex. The complexity is probably O(n!) or even exponential. If you try to parse every attribute of your crush, from the way they chew, to their music taste, to their political views... your brain’s processor will overheat. You will get hit with Analysis Paralysis.

That is why I am introducing a concept I use myself: Weighted Sum Model.

The Love Algorithm: U(x)

In Operations Research, we often use the Weighted Sum Model to make multi-criteria decisions. For example: choosing a factory location, selecting a server vendor, or in this case, choosing a potential wife/husband.

The concept is simple: We can’t have everything (that’s the principle of Pareto Optimality). We have to trade-off. But the trade-off must be measurable, not just “well, I guess I’ll accept it.”

The basic formula, if written in Python code, looks something like this:

def calculate_utility(candidate):
    utility = 0
    for trait in traits:
        utility += (trait.weight * candidate.score[trait])
    return utility + epsilon

Where:

U(x): The Utility Value (Eligibility Score) for someone to enter your life.
w_i: Importance Weight (how crucial criterion i is for you, the total must be 100% or 1.0).
x_i: That person’s score in criterion i (scale 1-10).
epsilon (Epsilon): Error Term or Noise Variable.

Now, one thing to remember: The variables and weights are up to you.

You want to use 7 to 8 variables because you are very detailed? Go ahead. Or are you the simple type who only cares about 2 main variables (e.g., “Rich” and “Breathing”)? Whatever works, that’s the user’s prerogative.

But personally, after performing feature selection on the thousands of complexities of a human being, I simplified it down to 5 Main Variables that I think are most crucial for long-term system stability. This isn’t about looking for “perfect”, it’s about looking for “optimal”.

Let’s break them down one by one.

x_1: Intellect (Weight: 40%)

This is the variable with the largest weight. Absolute. 40%.

Why? Because talking to someone whose logic doesn’t work is like trying to debug code without documentation: exhausting and frustrating.

Intellect here doesn’t mean they have to be a Cum Laude graduate or understand Advanced Calculus. No. This is about Logical Reasoning.

Can they distinguish between correlation and causation?
Can they have a discussion without using ad hominem arguments?
Do they believe in zodiacs or family group chat hoaxes?

Imagine you want to vent about a complex office problem, but their response is just, “Ah, that must be because Mercury is in retrograde.” Get out. Force Close. kill -9.

Life is full of problems that require problem solving. I need a partner who can be a co-pilot, not a passenger who just screams in panic when there is turbulence. If they can’t think rationally, your relationship will just be filled with unnecessary drama. Inefficient.

x_2: Humor (Weight: 20%)

Life is tragic. As Camus said, life is absurd. The only way to survive without going crazy is by laughing at the stupidity of this world.

That’s why Humor gets a 20% weight. But not just any humor. I need compatibility in comedy style.

If I complain using sarcasm (”Wow, the government is great, their solution is so out of the box that the box is missing”), they should laugh or add more sarcasm. If they get offended or think I’m being negative, system crash.

Humor is the error handling mechanism in a relationship. When you’re fighting, when you’re broke, when you’re sick, the ability to laugh together is the most powerful recovery feature. A rigid partner is like a server that doesn’t have an auto-restart feature. Once it’s down, it’s down forever.

x_3: Life Vision (Weight: 20%)

This is about roadmap compatibility. Where do you want to be in 5 years? 10 years?

If you want to be a digital nomad traveling the world, while they aspire to be a Civil Servant staying in one district for the rest of their life, that is called Divergent Vectors.

Vector A goes right, Vector B goes left. The resultant is zero. Or it tears in the middle.

Many people break up halfway not because they don’t love each other, but because their roadmaps are different. And foolishly, they only check this roadmap after 3 years of dating. That is a waste of compute resource. Check it at the start. If the visions are far apart and there is no meeting point (empty intersection set), abort mission.

x_4: Empathy (Weight: 10%)

This is a control variable to detect narcissism and psychopathy. Only 10%? Why so little? Because in my opinion, if x_1 (Intellect) is high, empathy usually follows (rationality demands we be fair). But we still need an explicit variable.

How do they treat servers? How do they deal with stray cats? If she is pretty, smart, but kicks a cat asking for food... that is a red flag as big as a political party banner on the roadside.

Empathy is the firewall that prevents you from getting hurt emotionally. Without empathy, your relationship is just a transaction.

x_5: Physical Aesthetics (Weight: 10%)

“Ew, why is physical appearance only 10%? You hypocrite!” Listen first. I’m not saying looks aren’t important. I’m saying the weight is small compared to the others.

Basic human biology does look for facial symmetry and healthy body ratios. That is hardcoded in our DNA for species survival. I admit, I also like looking at cute girls. But, physique is the variable with the Highest Depreciation.

Beauty/handsomeness is an asset whose value continues to drop over time (Time Decay). Meanwhile, Intellect (x_1) and Humor (x_2) are assets that can compound.

Do you want to invest in an asset that drops in value, or one that rises? Besides, if x_1 through x_4 are already high, usually x_5 becomes a positive bias. Smart and funny people automatically look more attractive. That is a beneficial halo effect.

Upgrade: When to Use WSM, When to Use AHP?

One important note before you start making a spreadsheet. This Weighted Sum Model is a tool that is naturally for high-level screening. Roughly speaking, this is the phase one filter.

It is very suitable for blind dates, acquaintances from green/orange apps, or meetings based on just 1-2 times getting coffee in a coffee shop where the music is too loud. In this phase, your data is still limited, the sample size is small, so you need a quick and dirty tool to determine: continue or skip.

But...

If you already have serious intentions, want to validate a deeper relationship so you don’t go bankrupt on emotional investment, WSM starts to feel insufficient. The subjective bias is still too big.

At this level, you need AHP (Analytic Hierarchy Process).

This is a different monster. AHP is a very structured decision-making method. Here, you don’t just give scores arbitrarily. You are forced to do a Pairwise Comparison... comparing every criterion head-to-head.

The questions become sharper: “Which is more important: Humor or Life Vision?” “If Vision is more important, how absolute is the difference? Scale 1 to 9?”

Later from there, a comparison matrix is created, the eigenvector is calculated to get precise priority weights, and the craziest part: there is a consistency test (Consistency Ratio). If your ratio is above 0.1, it means your answers are inconsistent (unstable), and you have to fill it out again. AHP forces you to be honest with yourself mathematically.

But then I thought again while sipping sachet coffee... c’mon, does a relationship have to be analyzed that deeply? Are we looking for a soulmate or creating a government project tender document?

The Epsilon: The Chaos Variable

Now, this is the most interesting part. You’ve calculated with advanced formulas, weighed the weights... yet there is still epsilon.

The Error Term.

This is the statistical trash bin. The place to dump everything that cannot be defined by my rigid logic. Her clumsiness spilling coffee. Her obsession with a Korean boyband I don’t understand. The weird way she laughs. The scent of her perfume that somehow makes me calm.

In an ideal mathematical model, we want epsilon to approach zero. We want 100% accurate prediction. But in human relationships, it is precisely this epsilon that is often the deciding factor.

Sometimes, you meet someone whose total score is only 7.5 (below the 8.0 threshold). Logically, they should be rejected. But the epsilon is massive. There is “something” that makes no sense that makes you comfortable.

Will I break my own rules for the sake of epsilon? A good data scientist would say: “Check first, is this an outlier or an anomaly?” But a human might say: “Screw the data, let’s run the risks.“

Many rational people struggle with this. They want everything measured, but they fall in love with a girl who is pure Chaos. A girl who is the personification of epsilon.

Conclusion: Don’t Forget to “Compile”

This Weighted Sum Model method is not God’s law. It is just a tool. A Decision Support System. Its use is so you aren’t completely blind when falling in love. So you have a logical checklist to hit the brakes when the dopamine is going full throttle.

My advice:

Determine the weights (w_i) your own version. Maybe for you, Looks are 50%? Go ahead, as long as you are aware of the risks (asset depreciation).
Do the scoring objectively. Don’t cheat by adding points just because she just chatted “Good morning”.
Set a Threshold (T_a). If the score is below 7, don’t force it. Don’t make it a charity project.

Life is too short to debug a relationship where the architecture was wrong from the start. Find someone compatible, optimize at runtime, and enjoy the process.

And if you find someone with a high score AND an epsilon that warms your heart... keep them. That is called a Global Maxima. Rare to find.

Happy optimizing.

Love is like Recursion without a Base Case. If you aren't careful, you will get trapped in an Infinite Loop until Stack Overflow…