The Wow! Signal After 45 Years

(waves crashing)

(gentle instrumental music)

- 45 Years ago,

on a sultry, summer evening in Delaware, Ohio,

something remarkable happened.

From the direction of the Sagittarius constellation,

an intense and continuous narrow band radio wave

bounced off a 30 meter tall,

flat reflector tilted towards the sky.

It then bounced off another parabolic reflector,

funneling it towards one of two detection horns

that received and amplified the signal.

Within the nearby control room,

there were no scientists or engineers to watch the moment

like in a Hollywood film.

Instead within that dark and lonely space,

the only sound was out of a busy line printer

and IBM 1130 computer,

diligently worrying and chirping away into the night.

The computer printout fell into the tray,

folding into the way to the subsequent pages.

But upon that page,

the irrefutable evidence

of the radio signal had been immortalized.

Sometime later,

a technician dropped off the printouts

at the home of volunteer, Jerry Ehman

who had agreed to look through them

for anything interesting.

scouring through the pages,

Ehman would carefully scan through the various numbers

with his red pen.

Most of the time,

the numbers were small images

representing the signal-to-noise ratio

the detectors received.

Numbers greater than nine,

had to be represented as letters for 10,

B for 11 and so on.

And so when Ehman stumbled across the sequence 6EQUJ5,

he realized that he was looking

at the highest signal-to-noise event he'd ever seen.

Instinctively, he circled the sequence

and wrote the now historic scribble, Wow.

That happened in 1977 and despite four and a half decades

that have passed since,

the Wow signal remains the most compelling candidate

for an alien radio transmission we have ever received.

I've long been fascinated with the Wow signal,

like an irresistible itch.

I've often found myself dwelling on it at night

in quiet moments,

wondering what was it?

Who sent it?

And what else could be out there?

Radio signals are detected from the sky all the time

to such an extent that there's an entire field

of astronomy dedicated to mapping the sky in the radio.

But the Wow signal found by the Big Ear Observatory stands

out as being incredibly exciting

because of four observational facts.

Number one, it's high signal-to-noise;

Two, it's particular frequency,

three, it's narrow band signature,

and four, it's apparently continuous nature.

For fact one,

the high signal-to-noise is crucial

because it removes any doubt that this could be some kind

of random statistical fluctuation.

The peak intensity of the Wow signal was 30 times higher

than the ambient background.

So that means that the probability of a random fluctuation

doing this was one intent to the power of 196.

That's an astronomically tiny number.

The bigger observatory could run for a trillion years

and never expect to see a fluctuation like that.

Yet more the moments just before

and after the peak

also show exceptionally high signal-to-noise values.

So there's no way that we can legitimately ignore this

as some kind of statistical fluke.

For fact two,

the signal frequency was almost precisely equal

to the famous frequency of 1420MHz,

also known as the hydrogen line.

This is a special channel long hypothesized

to be a likely wavelength that aliens might use

to attempt to communicate with us

because of two major reasons.

First, the cosmos is naturally quietest

at this part of the spectrum.

So there's less noise to interfere with the communications.

And second,

the hydrogen line corresponds to the emission caused

by the most basic electronic transition

within this simplest atom in the universe.

Anyone in the cosmos who understands quantum physics

should know about this special frequency.

And so it serves as a Schelling point,

a rallying point in frequency space.

Now, it should be noted that the frequency

actually wasn't quite exactly equal to the Hydrogen line,

but instead was 50 kilohertz above it.

However, this slight difference is consistent

with a variation expected

due to the Doppler shift of a transmitter

that is moving towards the Earth

at a modest speed of order of 10 kilometers per second,

which is a fairly typical velocity for nearby stars.

For fact, number three,

we turned to the narrow band nature of the Wow signal,

but how do we know this?

While the Big Ear Observatory had 50 detector channels,

each one with a bandwidth of 10kHz.

The Wow signal was only recorded in one channel,

channel number two of the 50 possibilities,

which means that its power is limited

to within a 10kHz window or even possibly much smaller.

So given the frequency of 1420Mhz,

that means that the bandwidth divided by the frequency

is less than seven parts per million, incredibly narrow.

Now in nature,

we simply don't know of any radio sources that do this.

For example, the Crab pulsar is often referred to

as a natural narrow band radio source,

but there, the spectral ratio is one part in 10.

Even if we turn to the recently discovered,

and mysterious fast radio bursts,

you still only get down to one part in 20.

The truth is that short of an artificial transmitter,

we really don't know how such a narrow band signal

could ever be produced.

Of the four facts,

this is perhaps the most important.

Finally, turning to fact four,

this signal appeared to be continuous during the 72 seconds

that it was observed for.

Now, when you actually turn the numbers

and letters into a graph, you see a bell curve like this,

not a flat continuous line.

But here's the thing

because of the Earth's rotation,

the observatory is sweeping over the field during this time.

Meaning that, in fact,

the bell like curve is exactly what you'd expect

for a continuous source.

Fact number four is important

because most satellites reside in a low Earth orbit

where they would pass through the Wow field much faster

than the 72 seconds that the signal was observed for.

For example,

the International Space Station would pass

through the Wow field in less than a second.

So you'd only get one single high digit,

not the persistent sequence of six that was observed.

These four properties are essentially exactly

what you'd expect from an alien transmitter

and likely a very high powered one.

Now, it has to be said that a human made orbiting satellite

cannot be excluded,

but they would need to be on a wide orbit

beyond the geosynchronous orbital altitude

to yield a sufficiently slow speed,

which was rather unusual in 1977.

Yet more that signal

will be utilizing a protected frequency,

since the hydrogen line is internationally agreed

to be a protected channel.

Spy satellites also don't make a lot of sense here

because after all radio astronomers

across the world are very closely monitoring

that wavelength.

So it really wouldn't be very secretive to be using it.

Emission from a comet has also been previously suggested

and made the news a few years back,

but it has largely been rejected by the community,

since they do not emit strongly enough at this frequency,

nor were the cited comets,

even in the beam at the correct time.

So in summary,

we have these four observational facts

that together look quite promising

for the validity of the Wow signal.

But on either hand,

we also have four holes in our knowledge that frustrate us

and ultimately lead the door open

to skepticism about the Wow signal.

So let's go through these four.

Number one,

we don't know if the signal was modulated or not.

Two, we don't know where it even came from.

Three, the signal was strangely only detected

in one horn.

And four, we've never seen it repeat, ever.

Points one and two are really just a product

of the observatory's limitations.

The detect was not sensitive enough

to detect AM or FM modulation.

So there could very well have been a message embedded within

the Wow signal that we just never saw.

Obviously, if we had seen something like that,

it would've removed any doubt that this was truly

from an artificial transmitter.

Point two,

the unknown location surprises many.

Surely a telescope knows what star is looking at, right?

Well, Big Ear has two big issues here.

First, each horn spans a fairly wide field of view

about one by 20 arc minutes on the sky.

Second, we don't actually know

which horn detected this signal in the first place,

that was never recorded.

Albert Caballero recently investigated this issue

showing here the large region of space

where the source could have come from.

There's thousands of stars in here,

but Caballero tries to narrow this down

to 60 near sun-like stars as the best candidate.

But frankly, we have no idea

whether the source was indeed a sun-like star

or even from a star at all.

Point three is a weird one.

The two detect horns actually sweep over the same patch

of the sky,

but their physical separation means

that they see that patch three minutes apart in time.

Now, because we actually don't know

which of the two horns detected Wow,

that means there are two possible scenarios.

In scenario A,

the first horn detected the continuous radio mission

from the Wow field,

but the signal then switched off in the three minutes

between the horns.

Or in scenario B,

the signal was not present in the first horn,

but then it switched on in the three minutes

between the two horns.

Either way, though,

it seems rather fortuitous that the Big Ear just happened

to be looking at the right moment

to catch the switch over,

unless Wow is just constantly switching on and off.

And then to top it all off, we have point number four,

which is perhaps the most frustrating part

of this whole story,

because despite many efforts to reobserve the Wow field,

there have been no detections of a recurrence

of that signal.

With just one detection, the signal is tantalizing,

but not definitive.

So where does that leave us?

Well, definitive proof

that the signal was indeed astrophysical

and not some terrestrial interference or space debris

demands a repeat detection.

But on the other hand,

a lack of repetition also does not comprise definitive proof

that it was spurious either.

So we are left in no man's land.

Or are we?

As subscribers of our channel will know by now,

my approach to science

often takes a rigorous statistical approach

to such problems.

And so one might ask,

could the absence of any recurrence

of that Wow signal be used to exclude it

as an astrophysical source

and then just move on with our lives?

And so this is where our new research paper comes in,

where we take a deep statistical dive into the Wow signal

and aim to squeeze that data as hard as we can.

Many of you will know that we've discussed the Wow signal

before on this channel in our "Black Swans" video.

There, I discussed a different Cool Worlds research paper,

where I developed an (indistinct) model

for dealing with one-off events.

The model works very well when you have a single observatory

with constant settings,

but Wow has been pursued now

from many different observatory.

Yet more in that video,

we saw how the unknown duration of the Wow signal

causes a real headache

and different assumptions lead to very different answers

about its likelihood of recurring in the future.

And so after that last video, I wasn't satisfied.

I wanted to go back and do the Wow signal justice

to fix these issues and complete a final analysis.

I wanted to know,

I needed to know,

could the Wow signal still be a viable alien transmission,

even after all of those no results?

Whilst making that video,

I started corresponding with Robert Gray,

who literally wrote the book about the Wow signal.

perhaps more than anyone else Gray has pursued Wow

with a feverish curiosity, visiting the original site,

interviewing the team and studying the technical diagrams

of the observatory.

He even set up a 12 foot radio telescope in his backyard

to continuously monitor Wow from home.

Remarkably, even though he was not trained as an astronomer,

he was awarded telescope time

on the very large array in New Mexico to look for the Wow.

The first amateur astronomer to ever use the facility.

So after the video was posted,

we continued to correspond about the Wow signal

and started working on a new improved analysis.

So what is new here?

What is the big idea?

The basic premise of our approach is

to create a responsive emulator of the Big Ear Observatory

and indeed the other observatories

that observed the Wow field.

Back in 1985, Gray had visited Ohio State University

who operated the Big Ear,

and managed to reconstruct

the exact dates of each and every time Big Ear looked

at a Wow field.

Even though Wow was only seen once,

Big Ear actually observed that field on 122 unique days,

spanning August, 1977 to March, 1984.

However, only 90 of these were actually useful observations

due to various issues at the site.

We also know that Big Ear has this peculiar observing mode

where two horns pass over the same patch

of the sky each day,

separated by three minutes

and each one observing for just 72 seconds.

So equipped with this information,

can now simulate a hypothetical alien transmission

from the Wow field?

And ask in how many of those 180 observations

would we expect to detect it?

Now, this is where the fund begins,

because with that emulator,

we can change the properties

of our hypothetical alien transmissions

and see how would it affect the observations

that Big Ear would get.

So for example, if the signal repeated like clockwork,

once every hour

and stayed on for about 10 minutes,

then we would expect about 30 detections.

However, if the signal repeats say,

once every millennia, instead,

then we would get zero detections.

In this way,

we have a Big Ear sandbox, if you will,

where we can play God with different alien transmissions

and see what Big Ear would report each time.

Now, there's a certain degree of randomness here

because apart from the alien signal

is periodicity and duration,

it also has some arbitrary start time,

which we simply randomize in the simulations.

So for any given alien signal,

we actually generate thousands of different realizations

of it and then average them together at the end.

So the real point of all of this is

that we can now ask the question,

what kind of alien radio transmission do we have to generate

such that it matches the properties that Big Ear saw

for the Wow signal.

In other words, out of 180 attempts,

how do we get just one success?

Now, to make progress though,

we have to choose between one of two types of alien signal.

In case one, a repeat like clockwork

on a regular constant schedule.

So for example,

this could be an alien transmitter

that resides on a rotating exoplanet

or in case two, it repeats randomly.

Perhaps this might be an intentional leakage

from an advanced civilization that sporadically points

towards the Earth.

Perhaps even it could be an automated beacon

that deliberately follows a random schedule.

Since it turns out this would actually minimize the chance

of a signal being missed by a receiver on a planet

with a fixed rotation rate like the Earth.

Now, some of you might be thinking to yourself,

hold on, What about if the Wow signal was a one-off event,

not repeating at all,

but actually technically a one-off event

is still a repeating event

just where the repeat time equals infinity.

And so mathematically,

we don't have to change anything here.

It is fully encoded within our framework.

The first case of a periodic signal has been studied

extensively already, including by Gray himself.

The summary of this is that any periodicity below 14 hours

is completely excluded.

And even beyond that looks very unlikely.

So if we assume that alien transmitters have to be periodic,

well, there's really not much chance

of Wow being what we might have hoped it was.

In this vein,

The best way to save Wow is

to assume that it's not periodic,

but rather it follows some random schedule,

but even randomness can be characterized statistically.

If we assume the average number

of pings per unit time is unchanging,

then that means that the events

must follow the well known class on distribution.

What this essentially says is that

yes, the events are being generated by some unknown,

random generator,

but the properties of that random generator

are themselves not changing in time.

So now the Wow signal is characterized

by just two parameters.

The duration of the pulses, let's call that T,

and the average rate of pulses per unit time,

let's call that Lambda.

Equipped with our Big Ear emulator

and our random signal generator,

we are now finally ready to figure out

what are the properties of the Wow signal.

To do this,

we essentially create a dense 2D grid of possible duration,

T values on the X axis,

and Lambda values up on the Y axis,

and ask in each one of these simulations,

how often do we get a Wow like signal?

In other words, just one success from 180 observations.

You can see here,

the color coded map that we recover,

where the white region there leads to the highest odds.

In that region, the peak probability is in fact, 32.3%,

almost a 1/3.

So that means that the fact that we saw just one Wow signal

from the years of Big Ear observations

is actually not that weird.

It's a one in three outcome.

If the Wow signal has a T and Lambda in that region.

This is why statistics is so important

because let's compare that to our intuition

or, in fact, the intuition of Jerry Ehman,

who remember was the person who first discovered Wow,

back in 1977.

In a 1994 interview, Jerry said,

"We should have seen it again

when we looked for it 50 times."

But here,

we find that it's actually not particularly unusual

that Big Ear never saw it again,

even after 90 observing days.

So long as a Wow signal has the right T and Lambda.

But hold on, so far we've only used the original,

Big Ear observatory data,

but remember in the years that followed other observatories

have pursued the field and also come up empty-handed.

Surely, if we add in this extra data,

it will finally kill the Wow signal.

Well, using our emulator,

we can easily fold in this additional data

and see what we get.

Of the various follow up efforts of the Wow field.

Three, really dominate.

And we'll focus on those here.

In 1994, Robert Gray used

the 26 meter Harvard Smithsonian Radio Telescope

for a total of eight hours on target.

Then in 2002, gray was added again,

but now using

the University of Tasmania's 26 meter radio telescope

on target for 84 hours in total.

And finally, in 2020 Harvard now joined the hunt

and used the Allen Telescope Array

to collect more than 100 hours on the field.

So that means there's been more than 184 hours

of accumulated observations since the original Big Ear data.

And all of this new data has even higher sensitivity

than the original Big Ear observatory did.

So look, I get it.

It's easy to see why a director and observatory

would just shoot down any proposition

to keep going at this point.

But critically is that a rigorously justified position?

Using our emulator,

this triage of new observations

definitely reduces the overall probability

of a random Wow signal

matching all of the existing observations.

Whereas before our peak probability was 32.3%,

about a third, the best matching T and Lambda.

Now slide down to a probability of 1.8%.

So about one in 50. Now one in 50 is undeniably improbable,

but it's also hardly implausible.

It corresponds to 2.4 Sigma in statistical parlance.

But in science,

we would only normally consider a hypothesis rejected

if across the three Sigma threshold.

So that means that we would have to change this 1.8% number

to 0.3% to finally reject Wow.

In our paper,

we can use the ator to actually predict how much more data

is necessary to get to that point.

And the answer,

around 1500 hours of additional observing time

would be necessary

or 62 days of accumulated observations;

that might seem daunting,

but it could be achieved in a single year

with aggressive scheduling.

And the payoff would be a conclusive answer at last.

Either, we see, Wow, repeat and know it's the real deal

a frankly Earth shattering result,

or we see nothing

and can finally exclude it as an astrophysical source

to highest statistical probability.

Now, one thing I haven't revealed yet

are what are the most likely properties of the Wow signal.

I've told you those most likely properties have

a one in 50 chance of occurring,

but what are the associated duration T,

and mean repeat rate, Lambda,

that correspond to that most likely solution.

Coupling at emulator to an inference framework,

we can constrain that the duration of the signal

is between 2 minutes and 37 hours to 95% confidence.

And as for the mean repeat rate,

that comes in between about once every other month

to 12 times per month, again, to 95% confidence.

Very slow repeat times,

those less than once per year

are strongly disfavored by the existing data,

as they almost always lead to zero total detections,

rather than the one Wow signal that was observed.

So in conclusion, Wow, is still alive, just about.

It may be a real, repeating signal,

a scenario with a probability of 1 in 50,

or it may be bogus after all.

Something that we could conclusively state

if we collected just two months extra of new data.

Here, we've tried to push the statistics as far as we can,

but there's only so much we can model.

Perhaps our assumption

that the mean repeat rate doesn't evolve is wrong.

Maybe someone is behind the controls of that transmitter

who has substantially modified the mean repeat rate

over the last 45 years.

Who knows why, but it's not impossible.

In science though,

we try to proceed with this simplest model possible.

that's justified by the data,

a philosophy that hearkens back to William Ockham.

And in this case with a single data,

I don't think we can easily propose a more elaborate model

in the absence of additional data,

but even so our analysis shows that hope remains.

Wow could be still out there just waiting for me

or for you to go to the telescope

and hear it again.

To close, I wanna share one last thing with you.

Robert Gray was the co-author

with me on this new research paper,

but during the writing up of this work,

he tragically passed away in December last year.

I didn't know Robert personally,

but professionally, he was a true gentleman,

with whom my interactions were always warm and spilling over

with enthusiasm for the Wow signal.

And that's clear from his life, the data analyst by day,

who trained himself to become the world's preeminent expert

on the Wow signal.

I strongly recommend you pick up this book.

This is "The Elusive Wow" by Robert Gray.

It is the definitive book on the topic

and a great read.

Look, despite the tragedy of Robert's passing,

I personally have some solace

that this final project he wrote on the Wow signal

led to the result that there is still some hope

of Wow recurring, not a lot of hope, but a little,

and maybe a Nerf that someone out there,

perhaps watching this right now might pick this up

and one day resolve the Wow mystery.

I'm glad that I got to help Robert with that.

So, until next time, stay thoughtful and stay curious.

(gentle guitar music)

- Why care about life elsewhere?

That's a good question.

Life here is riot of different forms, colors, sizes, noises,

environments, behaviors.

And I think that people probably have an innate interest

in whether something like that happened elsewhere.

And the only way to settle that question, of course,

is to go look.

(ethereal electronic music)