You have heard the advice, probably more than once. Your next job will not come from your close friends. It will come from a weak tie: the acquaintance, the person on the edge of your circle, the one who knows things you do not.
It is one of the few pieces of career advice with a genuine research pedigree behind it. It is also, as usually repeated, wrong in a way that changes what you should do about it.
Where the idea came from
In 1973, Mark Granovetter published The Strength of Weak Ties, now among the most cited papers in the social sciences. He had surveyed 282 men in the Boston area about how they found their jobs, and the pattern ran against intuition: the useful contact was usually not a close friend.
The logic is clean, and it is why the finding stuck. Your close friends move in the circles you move in. They hear about the same openings, know the same people, read the same rooms. An acquaintance stands somewhere else and can see what you cannot. Information travels furthest across the loosest connections.
That is a claim about direction: weaker beats stronger, for this purpose. It says nothing about how far the pattern runs, because a survey cannot say. It can show that people who found jobs happened to have the ties they had. It cannot show which ties would have caused a job if you gave someone more of them.
Nobody could run that experiment in 1973. So the directional finding hardened, over fifty years of repetition, into a rule of thumb with a slope on it: weaker is better, so collect more distant contacts.
Somebody finally ran the experiment
In 2022, a team led by Karthik Rajkumar, with Guillaume Saint-Jacques, Iavor Bojinov, Erik Brynjolfsson and Sinan Aral, published a causal test in Science. Not a survey. A series of randomized experiments run through LinkedIn's "People You May Know" algorithm, varying the ties people were actually offered, across 20 million people, 2 billion new connections and 600,000 job changes over five years.
Randomization is the whole point. If you change what ties people form and then watch who changes jobs, you can say caused rather than was associated with.
The answer was not the slope everyone assumed. It was an inverted U. Weaker ties do transmit more opportunities than strong ones, up to a point, past which there are diminishing returns to weakness. The strongest effect came from moderately weak ties: in the study's terms, roughly ten mutual connections between two people, and contact that was infrequent rather than nonexistent.
Fifty years of advice had been pointing slightly past the target.
Worth naming the scope, since this post is partly about people repeating findings they have not checked: this is one platform, digital connections, and effects were strongest in digital industries. That is the trade you make for an experiment this size.
Why "collect more contacts" was the wrong lesson
If weakest were best, the strategy would be volume. Add everyone. Accept every request. A wider perimeter means more distant contacts, and more distant contacts means more chances.
The inverted U says that strategy tops out and then works against you. A contact so weak that they cannot say anything specific about you is not a bridge anywhere. They can forward a posting. They cannot answer the only question that matters once your name is in the room: what are they actually like to work with?
A moderately weak tie can. That is somebody who worked with you once, two roles ago. Who saw the migration you ran, or the quarter you rescued, or how you behaved when it went wrong. You have not spoken in eighteen months and you are not in each other's circles any more, which is precisely why they hear about things you do not.
The part nobody mentions
There is a difficulty with those relationships, and it is not social.
That person's memory of your work decays on the same curve yours does. They remember that you were good. They do not remember what you did. So when your name comes up, the answer is a warm generality (she was great, ran some big project, systems I think?), which is worth a fraction of what the truth was.
You cannot fix somebody else's memory. You can make it trivially easy to refresh: a specific, accurate account of what the two of you actually worked on, ready to send, that they will recognize and can repeat in their own words. That requires you to have kept it.
The networking advice was never wrong, exactly. It was aimed at the wrong variable. The useful question is not how many people do I know. It is how many of them could say something specific about my work if somebody asked this week, and whether you could hand them the specifics if they could not.
This is what Standing Proof holds: what you actually did, in enough detail to hand to somebody willing to vouch for you. Write down something you finished recently, or read the method first.