All articles ←

Similar or complementary? One question, two opposite answers

10 September 2026·8 min read·By The Zugot Project team

"Opposites attract" and "like marries like" are two proverbs that contradict each other, and both are true. They are simply talking about different things — and once you separate the two kinds of dimension, the contradiction disappears.

The question that can't be answered

"Should I be looking for someone similar to me, or someone who complements me?" is probably the most common question people in the matchmaking process ask, and as phrased it cannot be answered. The reason is that it treats "a person" as one thing, when in practice a person is a bundle of entirely distinct dimensions — and each dimension behaves differently.

Break the question apart and two families of dimensions emerge, pulling in opposite directions. In the first, more similarity is always better. In the second, more similarity can be destructive precisely when it is perfect.

Family one: values, lifestyle and identity — maximum similarity

Worldview, level of observance, attitude to Torah study, degree of engagement with the modern world, financial priorities, ethnic background, what Friday evening looks like in the home. On these, the homogamy literature is unambiguous: the more alike the partners, the higher the satisfaction and the lower the probability of chronic conflict.

The logic is straightforward. These aren't character traits — they're rulings. Each of them generates dozens of small decisions a week: where the children are sent, what Shabbat lunch looks like, what happens over the holidays, what money is spent on. When both sides rule the same way, those decisions never come up for discussion. When they rule differently, each one is a small negotiation, and that negotiation repeats every week for decades.

One design detail deserves attention: these items aren't rated, they're picked. Asking someone to rate "how conservative are you, 1 to 5" is a high-friction question with a weak signal — people don't calibrate their identity on a scale, they recognise it. A label someone applies to themselves is far more reliable information than a number they invented to answer a question.

Family two: innate traits — complementary balance

Here the picture inverts. These are behavioural traits a person was born with and did not choose: extraversion, anxiety, warmth, a tendency to criticise, orderliness. On these, maximum similarity is not the goal — and at the extremes it is genuinely dangerous.

The professional insight behind this rule, which came from consultation with Michal Woshtein, is that once a person sits at the end of a scale, a partner who matches them at that same end amplifies the trait rather than balancing it. Two highly anxious people feed each other's anxiety: every worry of one confirms the worry of the other, and there is nobody to say "hold on, this isn't as bad as it sounds." Two very closed-off people will never start a conversation — each waits for the other to open, and genuinely neither does.

What an extreme needs is an anchor at the other end. A highly anxious person needs someone calm. A very closed person needs someone who talks. This isn't romantic opposites-attract — it's correction of a bias. A calm partner isn't "more interesting" to an anxious person; they simply return the couple to the middle.

The exception inside the exception: orderliness

One trait sits conceptually in the second family — it is innate, not chosen — and behaves like the first. Orderliness.

A very disorganised person and a very meticulous one do not balance each other. They grate on each other daily, over exactly the same things, endlessly. The sink, the laundry, what time you leave the house. There's no anchoring here — there is daily friction that manufactures resentment. So orderliness, unlike every other innate trait, is scored for similarity rather than balance.

Exclusion, not a points deduction

There is a substantive difference between two ways a system can treat a problematic pairing. It can score it low, or it can refuse to allow it at all.

The hard rules here — never pairing two highly anxious people, for instance — were framed from the outset as prohibitions, and so they are implemented as prohibitions rather than penalties. A pairing that trips one doesn't receive a low score; it never reaches the scoring stage. The reason is practical: a low score can be compensated for. If a couple accumulates enough points on values, interests and age, an anxiety penalty is simply absorbed and the pairing climbs back to the top of the list — precisely the pairing we set out to prevent.

So the scoring function only ever receives pairings that have already cleared every prohibition, and its job is to rank among them. It contains no disqualifying penalties, because nothing disqualifying reaches it.

What this looks like in the actual score

The weight distribution shows both families side by side. Similarity of religious and values outlook carries the highest single weight — sixteen percent — and beside it sit homogamy on lifestyle and identity tags, and closeness on religious stream and prayer rite. All of these reward similarity.

Opposite them stands the complementary-balance term on innate traits, at thirteen percent, rewarding exactly the reverse. And separate from both, an emotional-stability term that rewards both partners for being stable — not for being alike.

The practical conclusion for anyone in the process: when you ask yourself whether a candidate "suits you", split the question. Do we see life the same way? — here similarity is a good sign, and any gap is something to examine seriously. Is our character identical? — here total sameness at an extreme is a warning sign, and the right question isn't whether we're the same, but whether they steady me when I'm stressed.