These are confusing times for senior scientists in India. Those who read the Hindustan Times would last month have seen an encouraging message from the science and technology ministry. A draft note sent to all government ministries, the newspaper said, proposed raising the retirement age of government-employed researchers from 60 to 65. Scientists beyond 60, it said, are still productive and contribute to the scientific wealth of the nation. Most encouragingly, it claimed, the global average age of “top scientists” is 70.

Yet, Indian Prime Minister Narendra Modi seems to have a different agenda. Late last year, Modi’s office refused to grant permission for four lab directors at the Defence Research and Development Organisation to have their contracts extended past the usual retirement age. Modi, commentators say, wants to encourage young blood and fresh talent. He wants five of the research organization’s laboratories — including those that work on metallurgy, lasers and cryptography — to be headed by scientists aged 35 or under. At present, many young researchers go elsewhere, discouraged by the lack of opportunities in an organization dominated by the older generation.

Such bench-blocking is a problem for scientific organizations across the world. Last week, Nature reported on an initiative from the US National Institutes of Health (NIH), which for years has watched the average age of its grantees creep upwards (see Nature 518,146–147; 2015). The NIH has proposed a system of emeritus grants that will pay senior scientists to wrap up their research and close their labs, thereby freeing up money for the next generation.

If the young scientists waiting for their turn at the top table are growing impatient, then a study suggests that they have a strong case. As we report on page 283, analysis of some 20 million biomedical papers published over the past 70 years suggests that younger researchers are more likely than older researchers to be working on innovative topics. Out with the old? Not so fast: if you are good enough then you are old enough, certainly. But the latest analysis also suggests that the most productive groups teamed a young researcher with an old(er) hand. There is an age-old problem here, but it is not necessarily old age.