Fourteen years and more than US$1 billion after development began, the Stratospheric Observatory for Infrared Astronomy (SOFIA) has finally taken its 'first-light' science flight (see page 428). There seems every reason to think that SOFIA — a German-built 2.5-metre telescope fitted into a NASA-modified Boeing 747 jet — will be able to do excellent science. The infrared wavelengths that are its speciality are rich in information about a host of phenomena, including planetary atmospheres and star formation. However, per hour of observation (the metric that matters in astronomy), SOFIA will be almost as costly as the Hubble Space Telescope — the most expensive mission in the history of astronomy.

Will that science be worth the cost? This is not a question that usually gets asked about a brand-new observatory. Yet many astronomers are asking it about SOFIA. Partly this reflects their frustration with a project that has suffered repeated delays from its originally scheduled completion date of 2001, and a near-tripling of its originally projected development cost of $265 million ($360 million today).

Some of that frustration may well dissipate once SOFIA begins regular observing runs in November. But the question also reflects a more fundamental concern, which is that SOFIA suffers from much of the cost and complexity of space missions, without enjoying the full benefits of being in space. Although it is true that the high-flying telescope will get above most of the atmosphere's infrared-blocking water vapour, and will therefore have a much clearer view of the infrared Universe than any ground-based counterpart, it will not get above all of the water vapour. Its infrared vision will never be as clear as that of orbital telescopes such as the European Space Agency's Herschel Space Observatory, launched in May 2009. Nonetheless, SOFIA will cost some $100 million a year to operate over a projected 20-year lifetime, for a total of roughly $2 billion.

Could that money be better spent in more scientifically productive missions?

Could that money be better spent in more scientifically productive missions? A short-sighted answer is that, having spent so much time and money getting SOFIA into the air, it would be absurd — as well as politically embarrassing for NASA and the German Aerospace Centre (DLR), which is contributing 20% of the funding — not to get the maximum scientific return from the facility. It would also feel like a betrayal of the many scientists and engineers who have committed years of their lives to the project.

A longer-term perspective requires that NASA as well as SOFIA should be under the spotlight. The agency is notorious for delays and for overruns on everything from the International Space Station to the Constellation rocket programme. A report released last week by the US National Academies (http://go.nature.com/Ky7f8P) proposes some well-targeted ways in which NASA should clean up its act.

One priority high on the list: the agency should insist on independent cost and schedule estimates when science projects are being proposed and debated, instead of relying on estimates by advocates who have every incentive to be overly optimistic. NASA should likewise strengthen its ability to track cost overruns — which is surprisingly difficult to do at present, given that cost estimates for the various projects are wildly inconsistent about whether they include factors such as early development phases, launch costs and mission operations. The baseline estimate in every case should be a mission's total lifetime cost. The report also urges NASA to improve its system of technical progress reviews, which can sometimes proliferate to the point of disrupting progress.

But the key problem with SOFIA lies in astronomers' past inability to rank it alongside their other priorities and decide appropriately on its fate. Because it was already a mission under development, SOFIA escaped the prioritizing scrutiny of the most recent decadal reviews of astronomy. The lack of intermediate reviews minimized the ability to bring the project to a halt, as many would now argue should have happened.

As NASA, the DLR and astronomers address SOFIA's future, they should also develop tough and potentially terminal processes for reevaluating missions as costs spiral upwards.