The Matthew effect

A reader's summary of the idea that advantage breeds advantage: where the name comes from, what the sociologist who coined it was actually describing, the simple models that reproduce it, why platforms and rankings make it stronger, and where the phrase gets stretched too far.
The term at a glance
The Matthew effect is the tendency for those who already have status, resources or attention to gain more of them, simply because they already have them. It is also called cumulative advantage and, in network science, preferential attachment. Online it is the short answer to a familiar complaint: why the account with a million followers gets a thousand replies to a mediocre post while a careful thread from a new account gets none. It is one mechanism behind the power laws that shape audiences, wealth and city sizes, and a close cousin of the halo effect and network effects.
Origin
The American sociologist Robert K. Merton introduced the term in a 1968 paper in Science, “The Matthew Effect in Science”. He took the name from a verse in the Gospel of Matthew, in the parable of the talents: “For unto every one that hath shall be given, and he shall have abundance: but from him that hath not shall be taken away even that which he hath” (Matthew 25:29). In the parable the verse is about using what one has been given; Merton borrowed its wording for a much more worldly pattern.
His evidence came largely from interviews with Nobel laureates conducted by Harriet Zuckerman, his collaborator and later his wife. The laureates described how, once they were famous, they received disproportionate credit for joint work, while junior co-authors were overlooked; how papers were read because of the name at the top; and how two people who made the same discovery at the same time could end up with very unequal recognition. Merton later acknowledged that Zuckerman's research had shaped the idea more than the single byline suggested, which is, as he noted, an example of the effect itself.
History and context
The pattern had been modelled before it had a catchy name. In 1925 the statistician George Udny Yule described how the number of species per genus could follow a skewed distribution if new species tended to appear in genera that were already large. In 1955 Herbert Simon generalised the idea to word frequencies and city sizes. In 1976 Derek de Solla Price applied it to citations: papers that are already cited are more likely to be cited again. And in 1999 the physicists Albert-László Barabási and Réka Albert showed that a network in which new nodes link preferentially to well-connected ones produces the heavy-tailed link counts seen on the web.
The idea has since spread well beyond science. The education researcher Keith Stanovich used it in 1986 for reading: children who read well early read more, and so pull further ahead. Malcolm Gladwell's Outliers (2008) popularised the relative-age example, in which Canadian hockey players born early in the year were bigger at the age of selection, got better coaching and came to dominate elite teams. Economists use similar reasoning for wealth that grows faster than wages, and sociologists for neighborhoods and schools.
Main ideas
Recognition compounds
An early success makes the next one easier to win. Attention, money and trust are handed out partly on the basis of past attention, money and trust, so small initial gaps widen over time.
Credit flows to the known name
When a famous and an unknown person do similar work, the famous one tends to receive more of the credit. Merton saw this in co-authored papers and simultaneous discoveries.
Luck gets locked in
Cumulative advantage does not require the first winner to be the best. A lucky early break, fed back through rankings and reputation, can become a lasting lead.
Not only unfair
Merton also saw a function in it. A busy reader has to filter somehow, and a known name is a cheap signal. The cost is that equally good work by unknowns goes unseen.
The experiment that isolated it
It is hard to separate cumulative advantage from real differences in quality, since the winners might simply be better. A well-known 2006 study in Science by Matthew Salganik, Peter Dodds and Duncan Watts tried to do it. They built an online music market in which about 14,000 participants could listen to and download songs by unknown bands. Some participants saw how many times each song had been downloaded; others did not. The visitors who could see download counts were split into eight separate “worlds”, each evolving on its own.
When popularity was visible, success became far more unequal and far less predictable. The same song could be a hit in one world and middling in another. The best songs, judged by the group that saw no counts, rarely did badly and the worst rarely did well, but within those limits early luck decided a great deal. The study is often cited as clean evidence that social signals amplify small random differences into large, durable ones.
In the feed
Social platforms turn the Matthew effect from a tendency into software. Ranking systems promote posts that are already getting engagement, recommendation engines suggest accounts that already have many followers, and search results favor pages that already receive links and clicks. Each of these is reasonable on its own, because past popularity is a decent guess at future interest, and each feeds back into the next round. The result is the familiar shape described by the one percent rule and the creator economy: a small number of accounts collect most of the attention, and a long tail collects almost none.
The phrase now turns up whenever this feels unfair: the big creator whose recycled joke goes viral while the original poster is ignored, the bestselling author whose weaker book still tops the charts, the well-funded startup that raises again because it raised before. It is also used, more hopefully, in the other direction, as advice to build an early audience or habit because returns compound.
Critique
- Quality is not always noise. Some of what looks like cumulative advantage is ordinary merit. Famous scientists may be cited more because their work is better, and separating the two needs careful designs like the music-market study, which are rare.
- Advantages can fade. Fame, citations and market share decay without new input, and incumbents are displaced all the time. The models describe a tendency, not a law that the leaders always stay ahead.
- The opposite effect exists. Researchers have described Matilda effects, a term coined by the historian Margaret Rossiter in 1993 for women scientists whose work was credited to male colleagues, as a related but distinct pattern of credit flowing away from some groups. Advantage does not compound equally for everyone who has it.
- It is often a slogan. Online, the term is used for any inequality at all, which blurs a specific mechanism (early gains feeding later ones) with simple differences in luck, talent or starting wealth. The useful question is whether a feedback loop is actually present.
Impact
The Matthew effect gave researchers a vocabulary for something many institutions already did, and some have tried to counter it. Journals use double-blind review so that names do not decide what gets read; grant agencies have experimented with lotteries among qualified proposals; some platforms deliberately show new creators to a sample of users before popularity signals take over. None of these remove the effect, but they change where it starts. For a reader, the idea is a reminder to ask, when someone is described as the obvious leader, how much of the lead is the work and how much is the head start, and to remember that the answer is usually some of both.