Weak Ties: The Research Behind Why Distant Contacts Find You Jobs
The claim that acquaintances find you jobs more often than close friends do is one of the most repeated findings in social science, and one of the most casually mangled. The original 1973 paper made a narrower argument than the slogan suggests, and a large randomised experiment published in 2022 found the effect is real but not linear — the most useful contacts are not the most distant ones.
What Granovetter actually argued
Mark Granovetter’s “The Strength of Weak Ties” appeared in the American Journal of Sociology in May 1973 (volume 78, issue 6, pages 1360–1380). His definition of tie strength is worth quoting because people quietly substitute their own: a tie’s strength is “a (probably linear) combination of the amount of time, the emotional intensity, the intimacy (mutual confiding), and the reciprocal services which characterize the tie.” It is a bundle of behaviours, not a feeling.
The argument is structural, not sentimental. Granovetter’s central device is the forbidden triad: if A is strongly tied to both B and C, then B and C are very unlikely to have no tie at all. Strong ties cluster. Run that assumption through a network and you get a theorem that reads oddly at first — all bridges are weak ties. A bridge is the only path between two parts of a network, and a strong tie cannot be one, because the people at either end already share connections that route around it.
So weak ties are not better because acquaintances are more generous. They are better because they are the only edges that reach anywhere new.
The job-search numbers
The empirical hook in the 1973 paper is a study of professional, technical and managerial workers in a Boston suburb who had recently changed jobs, later set out in full in Granovetter’s 1974 book Getting a Job: A Study of Contacts and Careers. Of the 54 respondents who found their job through a personal contact and reported how often they saw that contact at the time:
- Often (at least twice a week): 16.7 per cent
- Occasionally (more than once a year but less than twice a week): 55.6 per cent
- Rarely (once a year or less): 27.8 per cent
Two things follow. The modal useful contact was already the middle category, not the most distant one — a detail the slogan drops. And the base is 54 people, in one occupational band, in one American suburb, half a century ago: a suggestive finding in support of a structural theory, not a law of career mechanics.
Why the structure pays: redundancy and brokerage
Ronald Burt’s “Structural Holes and Good Ideas” (American Journal of Sociology, 2004, volume 110, pages 349–399) tested the mechanism on something other than jobs. Burt surveyed 673 supply-chain managers at a large American electronics company, asking each one open question: what would you change to improve the firm’s supply-chain management? Senior executives then rated the resulting 455 ideas blind.
Managers whose discussion networks spanned organisational boundaries — low network constraint, in Burt’s measure — produced ideas the judges valued more, averaging a value rating of 2.4 against 1.5 for the most constrained. Fourteen per cent of the least constrained managers’ ideas were dismissed outright, against 43 per cent for the most constrained. Job rank, education and age added nothing once constraint was accounted for.
Burt’s phrase for this is a vision advantage: opinion and practice vary more between groups than within them, so people who sit between groups see options invisible inside either one. That is Granovetter’s information asymmetry, measured on idea quality instead of job leads.
The 2022 causal test, and the part that gets left out
Everything above is observational. In September 2022, Karthik Rajkumar, Guillaume Saint-Jacques, Iavor Bojinov, Erik Brynjolfsson and Sinan Aral published “A causal test of the strength of weak ties” in Science (volume 377, pages 1304–1310). They analysed a series of large randomised experiments on LinkedIn’s People You May Know recommender, which varied the prevalence of weak ties in the networks of over 20 million people across a five-year period, during which roughly two billion new ties, more than 70 million job applications and 600,000 new jobs were created.
The headline result supports Granovetter. Weaker ties increased job mobility more than stronger ones. But the paper’s own framing is that the relationship is not linear, and the nuance runs in two directions at once depending on which measure of tie strength you use.
| Measure of tie strength | What the experiments found | Practical reading |
|---|---|---|
| Number of mutual connections (structural) | Inverted U. Adding ties with more mutual friends raised the chance of a job transmission at first, but ties with more than about ten mutual friends in common reduced it. | Moderately weak ties — people who share a handful of your contacts — moved jobs best. Total strangers and close-cluster insiders both did worse. |
| Interaction intensity (messages, engagement) | Monotonic. The ties with the least interaction increased job transmission the most; the most-interacted-with ties increased it the least. | Rarely-contacted acquaintances outperformed people you message constantly. |
| Industry | Weak ties produced more job applications in industries with greater IT and software intensity, and in roles suited to remote work; strong ties did more work in less digitised sectors. | How much weak ties matter depends on what you do for a living. |
The inverted U is the part that almost never survives repetition. “Weak ties find you jobs” gets flattened into “collect strangers”, and the experiment says the opposite: ties with no shared context at all are not where the value peaked. What paid was the person plausibly connected to your world but not embedded in it — a former colleague two employers ago, someone you met once at a conference in an adjacent field, the person you shared a client with.
Be precise about what was measured, too. Job transmission here means a member applied for and took a job at a company where a new connection worked. That is a consequential outcome, but it is one platform’s version of a job search, and the authors themselves note the effect varies by sector.
What this changes about how you maintain contacts
The research supports a handful of concrete conclusions and does not support several popular ones.
Supported: breadth beats depth for information. Granovetter’s structural argument and Burt’s idea-quality data agree. A network reaching into several disconnected worlds is worth more, for job leads and for ideas, than a denser one inside a single world.
Supported: low-maintenance contacts are not dead weight. The least-interacted-with ties transmitted jobs most. A contact you speak to once a year is functioning as intended and does not need converting into a friendship.
Not supported: maximising connection count. The mutual-connection curve turns down beyond roughly ten shared contacts, and starts low at zero shared context. Indiscriminate connecting adds ties at the wrong end of the distribution.
Not supported: that weak ties matter equally everywhere. They did more in software-intensive and remote-friendly industries than in less digitised ones.
Practical maintenance, plainly labelled as craft
None of the following is measured. It is the routine that falls out of the findings.
- Keep a list, not a memory. Twenty to fifty names with a one-line note on what each person knows, where you met, and when you last spoke. A spreadsheet is enough.
- Prioritise the middle band. The people worth an annual message are ex-colleagues, ex-clients and one-time conference contacts — the ones who share some of your world but not your daily information stream.
- Make contact carry information. “Thought of you when I read this” with an actual link is a real message. “Just checking in” is not, and it costs you a small amount of credibility each time.
- Reconnect before you need something. The message that lands badly is the one whose first line in two years is a request.
- Watch for job changes. A contact moving to a new organisation is the single event that most changes what they can tell you, and it is the cheapest legitimate reason to get back in touch.
The practical version
The strength of weak ties is a real, causally tested effect and it is not a licence to collect strangers. The most productive contacts in the largest experiment ever run on this question were moderately weak: people with a handful of connections in common with you, contacted rarely, working somewhere adjacent to but not inside your current world. Build for reach across disconnected groups rather than for volume, keep a written list because you will not remember, and make each annual contact carry something the other person can use. If your network is dense, comfortable and entirely composed of people who already know each other, it is doing very little work.