SShortSingh.
Back to feed

LinkedIn experiment suggests mid-strength contacts drive more job moves

0
·1 views

Between 2015 and 2019, LinkedIn tested its People You May Know feature by randomly showing some members more distant acquaintances and others more people already in their circles, without informing them. A 2022 Science paper by researchers from LinkedIn, Harvard, Stanford and MIT reported that the experiments across more than 20 million members produced about 2 billion new connections and 600,000 new jobs. The findings pointed to contacts of middling closeness, rather than close friends or strangers, as most likely to lead to a new position. The result builds on Mark Granovetter's 1973 argument that acquaintances carry information that close friends often do not, a theory previously based on a small, US-only sample. The LinkedIn study was designed to show causation by randomly changing people's networks, rather than relying on correlations.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Product Operating Model Helps Teams Scale Without Losing Speed

As product organisations grow, informal coordination tends to break down and slow decisions. A product operating model addresses this by setting out clear outcomes, decision rights and shared evidence rather than adding layers of process. Marty Cagan and SVPG describe the approach as shifting how teams build, how they solve problems and how they choose which problems to pursue. Scaling pressures are often felt first in consistency, with many large product teams citing roadmap and process alignment as their main challenge. The article cautions against copying another company's model directly, noting that Spotify's squad structure was never fully implemented there.

0
ProgrammingDEV Community ·

AI-Native Tool Stack for Technical PMs in 2026, Organised by Layer

A new guide outlines a six-layer AI-native toolkit for technical product managers, spanning customer evidence capture, behaviour analytics, decision and specification tools, prototyping, coding agents, and a connecting layer. The central idea is that value comes from tools sharing context through protocols like MCP, so AI agents can work across them, rather than from simply counting AI features. The article cites surveys showing high AI adoption in product work, along with widespread developer distrust of AI accuracy, which argues for a review step in any pipeline. It estimates that a solo founder can run a core stack for roughly $20-65 a month, while a five-person team may spend a few hundred dollars, mostly on per-seat plans. Prices were reportedly checked on vendor sites in November 2026 and are expected to change.

0
ProgrammingDEV Community ·

Playwright MCP v0.0.72 renames browser_run_code to browser_run_code_unsafe

Microsoft released playwright-mcp v0.0.72 on April 30, 2026, renaming the browser_run_code tool to browser_run_code_unsafe to make its sandbox risks explicit. The change does not alter functionality, since the tool still executes arbitrary code within the Playwright browser context. The release also adds browser_network_request, which lets users fetch individual network requests by index, including headers and body, with an optional filename for direct file output. Teams that reference the old tool name in CI pipelines or MCP client configurations may encounter silent 'tool not found' errors and should update their references and security documentation. The release also fixes an unhandled rejection that previously crashed the server, which may change error patterns in logs.

0
ProgrammingDEV Community ·

OpenSSL CVE-2026-121481: Base64 encoder overflow affects 4.0.x and 4.1.0

A heap buffer overflow has been reported in OpenSSL's EVP_EncodeBlock() Base64 encoder, tracked as CVE-2026-121481. The CVE record gives it a CVSS 3.1 score of 9.8, though that score is reportedly not OpenSSL's own rating. The bug occurs because the function uses int lengths, so inputs of roughly 1.5 GiB or more produce output that overflows an int and can be written past the caller's buffer. The record lists OpenSSL 4.0.0 to 4.0.3 and 4.1.0 as affected, with fixes 4.0.4 and 4.1.1 reportedly scheduled for 1 December 2026. No public exploit, observed attacks, or CISA KEV listing had been found in the reports reviewed.

LinkedIn experiment suggests mid-strength contacts drive more job moves · ShortSingh