THE world’s most sensitive radio telescope was designed to catch a whisper from the dawn of time.
To build it, we switched on the one thing its whole existence was meant to keep out.
The nations that paid for a cathedral of quiet were never shown that it went up with no roof.
I know, because I helped put it there.
For nearly four years I led infrastructure construction on SKA-Low, the low frequency half of the Square Kilometre Array, in the Murchison, a stretch of Western Australian desert chosen for one quality: silence.
The site sits inside the world’s largest radio-quiet zone.
A mobile phone is contraband.
A spark plug is a threat.

When it is finished, SKA-Low will listen for hydrogen from the cosmic dawn, the first few hundred million years after the Big Bang, a signal so faint that one badly shielded laptop can drown it.
Now the line nobody at the top wants in print.
We spent years and a fortune guarding the quietest place on Earth, then lit it up with Starlink, the noise the astronomers had begged us to keep out.
SpaceX’s Starlink is the largest satellite constellation ever flown, thousands of spacecraft and still climbing.
A Curtin University study, using SKA-Low prototype stations on this same site, recorded 112,534 detections of 1,806 individual Starlink satellites in 29 days, at the very frequencies the telescope is built to hear.
Some of that leakage falls in bands protected for radio astronomy.
It does not come from the transmitters a regulator can switch off.
It comes from the electronics on board, unintended, and you cannot point a telescope away from a sky that fills up everywhere at once.
We did not fail to protect the sky.
We plugged into the thing compromising it, and we knew.
Why would anyone do that?
Because the plan we were handed had a hole in it you could drive a road train through.
Here is the detail that should stop a funding committee cold.
The program for a global science project of this scale, written and costed before my time, never planned where the workers would sleep.
The crews lived in temporary camps, but the camp was not in the program.
You cannot build an array of this size in the deep outback without one, and nobody had costed it.
So we found contingency that should never have been needed, for a camp that should have been in the plan from day one.

• Kaushik Bose
Once people live in a camp, they need to be connected.
The fibre, the obvious lifeline, was an afterthought, laid late.
A plan that forgets where the workers sleep has already forgotten the workers.
And forgetting the workers, in that country, is how people get hurt.
More than 200 of them, spread across a zone some 70 to 80 kilometres wide, planting antennas by hand.
Out there the light goes fast and the heat does not.
You manage hydration all day, and the ground is full of things that bite and sting.
A rolled ankle at the far edge of the array is not an inconvenience.
It is a man alone in the dark, kilometres from the nearest vehicle, in country that does not forgive being underestimated.
The processing facilities scattered across the array needed a line back to the central station and to Perth.
If a man went down, the way to reach him was a satellite phone or a Starlink link.
There were a couple of near misses while I was there.
Take that connection away and you do not lose email.
You lose the ability to find a hurt man before dark.
That was the real trade: the scientists’ silence against the workers’ safety.
Put that way, there was only one answer anyone could give, and I would give it again.
That is what makes it a trap and not a simple scandal.
Ant Schinkel, the construction director in my time, understood the astronomers better than most.
He described their fear the way an observer feels it.
You are watching the open sky, and a lot of tiny shining mirrors are shining back at you.
He also told me the truth of his own position.
He had no other choice.
He allowed the main contractor to use Starlink, and eventually, he said, he got used to it.
CSIRO, the landlord, ran a famously strict site, and yet its rules did not stop Starlink being used during construction hours.
The quiet zone, it turns out, was written to stop a farmer’s radio, not a megaconstellation.
Its rules control what is transmitted on the ground.
CSIRO’s own spectrum manager said it plainly this year: the zone was designed before the satellite era and offers no protection from above.
We built the walls very high and left the ceiling open.
But the loudest noise on that project was never in the sky. It was in the schedule.
The planner who built the program told me it had become a rolling stone, with no margin left in it.
The critical path kept moving.
Project controls, headed by Andrea Casson, kept pushing to re-baseline.
Re-baselining resets the official plan, so the new dates become the ones you are judged against.
Do it often enough and a project is never late, because the finish line walks backward as fast as you walk toward it.
The change-control process, meant to govern slippage, had become the means of changing things.
The watchdog was the escape hatch.
The planner’s own words, given to me privately, were sharper than anything I could write.
“They don’t even look into the schedule,” he told me.
“Only a few project managers know what’s existing in their scope.
“You are one of them. The others don’t know.”
He was also quietly angry that after he had built the program, someone else was brought in to make it look good for submission.
The plan the funders saw was the handsome one.
The real one was a mess he was no longer allowed to fix.
A schedule dressed for a board meeting is a costume.
In the evenings I was part of all of it.
I would come in off the array, boots red with dust, the desert going cold outside, and hook my laptop to the Starlink line at the site office to write up the day.
Some days I used the same link to test commissioning nodes.
I filed my reports over the very signal the scientists were fighting to keep out of their sky.
I never felt clever doing it.
I felt like a man drinking from the only tap in a drought, knowing whose water it was, and drinking anyway.
So let us stop pretending this is a story about greedy satellites.
The astronomers did not lose their quiet to Elon Musk alone.
They lost it to a plan that forgot the camp, forgot the fibre and forgot the people, then kept redrawing its own dates.
Starlink filled the hole the planning left open.
The Americans gave us the roof.
Their constellation kept the worksite connected and kept a hurt man findable in the dark, while it leaked into the sky we were building to hear.
China is a member of the observatory, one of the nations whose money, engineers and scientists stand behind the telescope.
China is also building Qianfan, its own megaconstellation, into the same sky.
And the rules that might restore some silence will be argued at the World Radiocommunication Conference in 2027, in Shanghai.
I did the one thing an engineer can do with a contradiction.
I costed it.
I ran a net present value analysis on what it would take to run the work, and the numbers pointed somewhere obvious.
We were an intergovernmental organisation, and China was one of its nations.
Bring a Chinese team in to support our Australian colleagues, I recommended to procurement.
It was cheaper, and it was within the rules.
A senior project manager gave me the answer.
He could not accept the NPV the way it was written, he said, because of the diplomatic reasons attached to it.
The arithmetic was fine.
It just pointed at the one partner nobody could name out loud, on a telescope built to hear the first light in the universe, under a roof rented from one superpower, in a silence the other is helping to fill.
• Kaushik Bose worked on the delivery of SKA-Low in Western Australia between 2022 and 2025. No confidential information has been used in this article.
by KAUSHIK BOSE