Somewhere in the house, a small plastic box is blinking behind a stack of unopened mail. The maths running inside it was honed on a hunt for exploding black holes.
Nobody ever caught one.
Hawking’s 1974 paper and the man who went looking
In March 1974, Stephen Hawking published a page and a half in Nature under the title Black hole explosions?, question mark and all. His argument was that black holes leak. Big ones leak so slowly that the universe is far too young for it to show. Very small ones, born in the chaos after the Big Bang, would evaporate fast and finish with a bang. Anything lighter than about a thousand billion kilograms would already be gone, Hawking calculated, releasing in its final tenth of a second roughly the energy of a million one-megaton hydrogen bombs.
The explosion itself would show up mostly as gamma rays, and the prospects for catching those looked bleak. Three years later, writing in Nature, Martin Rees floated a better target. On certain assumptions, electrons and positrons flung out of a dying hole might hit the galaxy’s magnetic field and throw off a radio pulse, exactly the sort of thing a radio telescope could hunt for. One paper, one set of assumptions, and suddenly the job belonged to radio astronomers.
John O’Sullivan was well placed to try. He finished his doctorate at the University of Sydney that same year, then left for the Netherlands and the Westerbork telescope. He stayed nine years and ended up running the group in charge of its receivers, according to the citation for his 2009 Prime Minister’s Prize for Science.
In 1978 he and two colleagues published their own search in Nature, and hedged it carefully. A radio pulse “may also be generated”, they wrote, with the energy estimates uncertain by many orders of magnitude and no evidence that primordial black holes exist at all.
The signal would have been pathetic in any case. Faint, buried in background hiss, and smeared on the way in, turned from a sharp spike into a rounded lump by the long trip across the galaxy. Pulling one out meant leaning on the fast Fourier transform, a method for tearing a messy signal into the individual frequencies hiding inside it, then reassembling it with the distortion stripped out. O’Sullivan liked the technique enough to have it cast in silicon, working with Austek Microsystems on a 160,000-transistor chip that ran the transform in real time for the Australia Telescope at Narrabri.
The bursts never showed up.
Why a living room is a hostile place for radio
The ambition, in O’Sullivan’s words to that same prize citation, was to “cut the network cable that linked every office computer”.
He was back in Australia by 1983, and by 1990 CSIRO, the Australian government’s science agency, was looking for ways to turn its radio physics into products. Indoors, wireless networks were slow and unreliable, and nothing anyone tried made them behave.
The culprit was multipath. When a US court later picked through the patent, it described the mechanism in flat language, quoted in CSIRO’s own history of the technology. Radio waves reflect off walls, furniture and whatever else is lying about, so a single signal reaches the receiver several times over, from several directions, each copy slightly out of step with the others.
Rooms ring, in other words. A transmission comes out the far end faint, echoing and smeared, which is a description O’Sullivan had run into somewhere before.
He and four CSIRO colleagues, Graham Daniels, John Deane, Diethelm Ostry and Terry Percival, worked out the fix. Data could be split across many frequencies and sent in parallel, then stitched back together at the receiver, with the same transform used to unpick the echoes.
The patent and the companies that would not pay
The document that fourteen of the world’s largest technology companies would later try to destroy was titled, with no flourish at all, Wireless LAN.
CSIRO applied for an Australian patent in 1992 and the US grant followed in January 1996. The ideas went into the 802.11a standard first, then into 802.11g and 802.11n. A spin-off called Radiata built the first chipset, and in November 2000 Cisco agreed to buy the company for stock worth about US$295 million, closing the deal the following February.
Then came the awkward part. Chipmakers shipping the technology declined to pay for it, so in 2005 CSIRO sued Buffalo Technology as a test case, and the industry’s heavyweights piled in to have the patent declared invalid. Hewlett-Packard broke ranks in 2009. As the National Museum of Australia tells it, the rest folded one by one, for a reported settlement above A$205 million.
The sequel was bigger. In 2012 the last holdouts settled as well, the big US mobile carriers among them, handing over more than A$220 million, as reported by the ABC. By that point CSIRO held licences with 23 companies covering most of the industry, and total revenue from the invention had passed A$430 million.
The patent expired the following year and the royalty tap closed. The technology, by then in billions of devices, stayed exactly where it was.
What the failed search was worth
So what does a dead end buy?
Nothing here was in the funding application. A government science agency did not set out to earn A$430 million in licensing revenue by chasing a theoretical object that may not exist. The exploding black hole was the goal, the signal processing was only the tool, and the tool proved worth vastly more than the target.
O’Sullivan went back to telescopes, joining the design team for the Australian Square Kilometre Array Pathfinder, an instrument built to look billions of years into the past.
The primordial black hole remains hypothetical. Half a century of searching has produced no detection, an unusual foundation for one of the more profitable accidents in the history of public research.
