These pages are concerned with developing and promoting a specific programme for finding extra-terrestrial intelligences through their attempted communications with other intelligences (such as ourselves).
A paper on the use of Pulsars as SETI Beacons is now published in the International Journal of Astrobiology (actually appeared in Spring 2004). It is co-authored with Dr Ian Stevens in the School of Physics and Astronomy at Birmingham University. It is an extension of the ideas sketched out below, and offers predictions of target stars for SETI efforts. You can find it here.
Currently - Mid-February 2005 - I am working at Arecibo with Paulo Freire doing SETI in line with the proposals in the published paper. You can see a couple of photos: here is the telescope and here is me in the driving seat.
There are several approaches to searching for ETI. One can do a broad survey
‘listening’ for anything that’s out there! One can do a targeted/directed
search. One can make assumptions about what to listen for:- perhaps ‘leakage’
from ET’s TV soaps or mobile phones, perhaps something broadcast with the
intention of being discovered – an omni-directional ‘beacon’ for example. One
can direct one’s search for specific reasons – known planetary systems around
distant stars, for example. More information about these various possibilities
can be found on the SETI League’s website.
SETI League, for example, run a project called Project Argus which
has as its aim a comprehensive all sky survey. In their words:
Project Argus is an effort to deploy and coordinate roughly 5,000 small radio telescopes around the world, in an all-sky survey for microwave signals of possible intelligent extra-terrestrial origin. When fully operational, Project Argus will provide the first ever continuous monitoring of the entire sky, in all directions in real time.
In my view this approach is inadequate because it is based on
a weak model of intentionality. The presumed intentionality in general survey
work is that ETIs will be broadcasting in the RF or visible spectrum, and will
be doing so omni-directionally. Our best
bet, therefore, is to search in survey mode, picking specific signal
frequencies for the survey on the basis of some simple assumptions but
otherwise not presuming much at all. The survey/search is conducted
omni-directionally – thus matching the presumed intentionality.
In the survey approach the model of intentionality is weak because it assumes
little more than a general intention to communicate – somehow or other, to
somewhere or other. The weak intentionality means that the search space is very
large, both physically and in terms of signal properties. NB – this
discussion of intentionality discounts surveys based on finding ‘leakage’ radio
emissions not designed for ET contact.
Stronger assumptions are possible which narrow down the search space and which
express a sense of sampling the sky. These are targeted surveys. One way to go
about this is to search for planetary systems around stars and then focus on
those stars with such systems as possible sources of broadcasts from ETIs. This
still requires a general survey to be done, but focussed on looking for likely
candidates identifiable by other physical properties than structured
transmissions. The survey might be being done for other reasons anyway, which
is efficient of resources. However, although the assumptions are stronger from
our perspective, they still assume that the ETIs’ broadcasts are
omni-directional and weakly intentional.
The whole enterprise of looking for ETIs should, in my view,
assume on their part an intention to communicate, and also that intentionality
is not unique to humans and that it is ‘readable’ or ‘interpretable’ in a
pan-specific way. If we assume such intentionality then we should assume that
the ETIs will do what can be done to be found – the point being that their
primary intention is indeed to be found! We should therefore attempt to match
their intentionality – what would we do in order to be found? What does it mean
to be found in the galaxy?
Simply signalling an existence, a presence, is not enough – omni-directional
‘beacon’ broadcasts are not helpful because they rely almost entirely on
content to uncover locational detail; detection of an artefactual signal merely
indicates presence of an ETI along the ‘line of sight’ from the receiver. Can
we do better, and thereby determine a way in which an ETI could do better? This
racks up the intentionality on both sides, which is what we should be
presuming. This way of reasoning means that if we can design a procedure which
we would use, and which is maximally expressive of our intention to be found,
then if we use that to define our search procedure and we find something
artefactual we can be reasonably sure of its intentionality.
If we were to broadcast in a directed manner, the fact of
selecting specific directions would have to be identifiable to an ETI.
Likewise, we would need to direct our search in directions specified in the
same way. This use of the sameness of
specifications is what would immediately flag shared intentionality. Note also
that directing transmissions (motivated sampling) can lead to more energetic
transmissions (directional RF antennae, for example). We should assume also
that consideration of energy efficiency, signal to noise ratios, and etc., will
constrain an ETI’s efforts as much as ours. Omni-directional broadcasts don’t
look optimal, directional broadcasts look better, but clearly putting all one’s
energy into one signal in one direction would be counter-productive. Are there
signal properties which could enhance signal to noise ratios? Yes - short,
sharp pulses will do the trick, but at what frequency, with what duty cycle,
and with what spectral properties?
Let us begin to answer these questions by looking at directionality. How could
we specify the directions in which to broadcast a signal in such a way that an
ETI could use the same specification to constrain a search, and thus in a way
which we could use to specify where we should look assuming they had
transmitted according to the same specification? Answer: we transmit towards
pulsars. And, toward each pulsar we transmit a signal with that pulsar’s
temporal properties. Our directed search is thus equally easy: we simply look
directly away from pulsars, and for signals with the temporal properties of
those pulsars. The basic signal is thus a pulse with appropriate temporal
properties. Averaging over many pulses will help drag the signal out of noise,
if it is not strong at the receiver. Its artefactuality is not lost through
such averaging. The spectral characteristics are discussed below.
Artefactuality is incontestible if such signals are found, and thus
intentionality is being exploited more forcefully than is the case on the
omni-directional survey model. Note that the distribution of pulsars means that
for each searcher/transmitter space is being sampled. But on the assumption
that there are large numbers of ETIs in the galaxy, and that both transmitters
and receivers will not have overly narrow beams, we can presume that sampling
will not diminish the chances of discovery too much. And it is discovery that
we are concerned about at this stage.
We can do more with our pulsar signals than merely look in the right places.
Some pulsars have changing pulse rates, and if this is mimicked also by an ETI
then distance to that ETI could perhaps be recovered directly. This would
provide a means to identify the location of the ETI without even decoding any
signal content! This is really the basic concept of being ‘found’ – being
located in the galaxy.
Could an ETI use a simple way to express location – a way we could use to
locate ourselves for others, and which we could use to recover their intention?
Yes – all we have to do is pick pulsars in different parts of the sky, and
indicate which they are (because we exist at the point ‘triangulated’ by the
lines from each pulsar, and assuming of course that we have the locations for
the pulsars reasonably well determined). There are two ways to do this which
are unambiguous, and either could be used. Sorting out which is which is not
difficult.
The right-angled system requires three directions to be specified, at
right-angles, one of which will be the line of sight to the detecting
intelligence. (The assumption here is that orthogonality is a basic concept.)
To signal location in this system one needs to transmit in the direction of one
pulsar information about the identity of the other two pulsars. This is
plausibly done by signal pulse rate ratios (pulsars are identified by pulse
rate, so rate ratios will be an obvious code). A binary coded number ‘p’can be
transmitted ‘q’ times to indicate the ratio p/q where q is the pulse rate along
the line of sight. For the second pulsar with rate ‘r’, r/q can be transmitted
the same way. Clearly, it is sensible to pick pulsars with manageable pulse
rate ratios, provided that the right-angledness of the directions can be
approximated well enough.
Binary codes are elemental in information theory, which could be assumed to be
known. Byte size would need to be recoverable irrespective of content – perhaps a 16 bit byte with a 17th pulse as
separator would be so artefactual and readily recovered that it would work.
Other proposals might make more sense – but they can all be tried on recorded
signals.
The tetrahedral axis system simply requires the signal source to be located at
the centre of an imaginary tetrahedron with pulsars in the directions specified
as from the centre to the four corners. The consequence is that along the line
of sight to one pulsar (and thus detector) information must be transmitted
about the other three pulsars. Recovery of information about three pulsars thus
indicates the system is being used (only two would identify the right-angled
scheme). The pulsars identified would, we presume, be independently
identifiable from their pulse rate properties, and locatable in space by the
detector, and thus the location of the transmitter would be found. Which is the
point.
In essence, therefore, the basic intentional message of the transmitting ETI
is: “We are here”. And the ‘here’ is recoverable absolutely. The survey
approach assumes less plausibly that an ETI is merely signalling: “We exist”.
Nothing has been said so far about the signal properties other
than rate and mark/space ratio (duty cycle). These mimic the pulsar towards
which the signal is being transmitted, and thus serve to provide the detector
with information for conducting the search. But what about the spectral
properties? This is rather unconstrained but should cover spectrally obvious
frequencies specified by obvious emissions, resonances, etc., and to which
detectors can be tuned. There need not be too many frequencies. Some will need
to be transmitted unvaryingly with every pulse – to be used to normalize pulse
strength in the detector. Others will need to be modulated on/off for binary
signalling, but with some redundancy and with careful selection for a number of
channels. Pulsar identification data could go out with high redundancy, perhaps
using just one channel. Other data can be more complexly coded. Other coding
schemes could be based on very small variations in timing of the pulses at
specific frequencies, which would have the benefit of maintaining signal
strength in averaging processes even if the pulse edges were slightly blurred.
Bytes could be transmitted in single pulses if 16 channels were available –
perhaps alongside the sequential 16+1 structure for location data (the latter
would be more immune to signal loss and/or inadequate signal strength).
To be more specific about one possibility – for illustration – the 1420MHz
Hydrogen line in the rf spectrum is often considered a basic signal to look
for. Fine – but why not look for artefactual frequencies based on that as well,
and simultaneously? So 1420, 2840, 4260, 5680, 7100, 8520.... up to the 17th.
Note - signals at these frequencies can be summed to specify a single
pseudo-pulsar signal for initial searching. Once found then more sensitive
detectors can be deployed on the parallel channels. The advantage of this
approach is that summations can be specified to yield maximum signal/noise
ratio – the known pulse rate means one can do the appropriate signal processing
to look deep into noise for pseudo-pulsar signals. Sure – the number of
parallel channels will need to be explored carefully – but a well-defined
strategy is not difficult to devise. For some thoughts on possibilities go here.
And even more elegantly – the search is terminating! There will quickly come a
time when one has exhaustively searched the plausible/candidate directions for
the plausible signal characteristics.
But that is deafeatist! We should assume it is worth thinking
about message design also. The message content (noting still that we are
concerned with identification/finding other ETIs, and with telling about
ourselves) should in part be limited to location information, conveyed
variously for redundancy (so, location information could also be sent
graphically, assuming perhaps a raster scan imaging system). Rasters
could be decoded if asymmetric images were sent – for example a graph of the
spectral output of the sun around which the ETI’s planet revolved – and note
this could provide an optical basis for locating the ETI also. Even
colour images of the inhabitants would be feasible (using markers on the same
spectral graph to indicate colour filters.....).
Although apparently fanciful the shift from using sequences of numbers to
specify location, to using byte strings for conveying images makes sense in
intentional terms. The transfer from uni-dimensional signal stream to parallel,
and to 2-dimensional representations, permits much more information about the transmitting ETI to be conveyed. This goes with
elaborating the original intention: “We are here” becomes “We, who look like
this, are here, which looks like this”. To be sure, the scanning in a raster
system is arbitrary, but rastering is an obvious way of turning a 2D image into
a stream of bytes (and vice versa). An alternatively obvious way is based on
radial/spiral scanning. Moving from centre to outer edge or vice versa (which
is being used is obvious from image properties) in either a clockwise or
anti-clockwise spiral will convey an image. Interestingly, the arbitrariness
remains a problem but is suceptible to a solution. The direction of scan in
either system is recoverable if an image includes a 2D depiction of the 3D
arrangement of the pulsars used for recovering location. The absolute
arrangement of the pulsars in space will map only onto one of the two
alternative representations. Why does this matter? – because it helps to flesh
out the “we look like this” detail by conveying absolute left/right information
(assumptions about up and down have to be made given that planetary mass will
provide gravitational grounding, and a distant sun will provide an illuminated
sky because of an atmosphere, and the details are available in images –
reasonable assumptions). The notion of seeing is also assumed in this scheme,
but the generality is that ‘eyes’ are 2D sensors, and these are plausibly
omnipresent (with appropriate arrangements for deriving 3D representations, as
we do). The plausibility is prompted by the fact of 4D space-time in which we
presume we and other ETIs are operating.
Having detected a “We, who look like this, are here, which
looks like this” message – how do we answer back? We reply in the same vein,
using the same techniques, but in an image of ourselves we hold up an image of
the “we” to whom we are replying. The signals will have to be transmitted
alongside real pulsar signals – but sharing only the temporal properties.
The pulses we send can be interleaved with the ‘real’ pulses.
Why didn’t we find such signals already, alongside the pulsar pulses/sources we
have mapped? Doesn’t the lack of such detection provide proof negative? No –
ETI replies would accompany pulsar pulses only if an ETI following this scheme
exists upstream (toward the pulsar) from us and is replying to an ETI existing
downstream from us who is also following this scheme – we would have to be
eavesdropping, but to do this the three intelligencies would have to exist in a
straight line lined up with a pulsar – an eventuality remote enough to be
assumed seriously improbable.
Aside from providing the essential “We, who look like this, are here, which
looks like this” intentionality in the reply, we could afford to be concerned
to convey other details. Such elaborations of content would have to be
designed with careful attention to intentionality – we would need to be assured
that the ‘why’ behind any message is plausibly recoverable. That is for another
day.
Comments by Jean Shneider, following my presentation in Paris at the workshop on message design (see below), indicate that the approach advocated here is indeed novel. Jean also pointed out that pulsars have a high proper motion, which means that their use as beacons comes with the need to adjust the direction in which one observes – an ETI’s alignment with a pulsar (from our perspective) will change significantly during the time it takes ETI’s messasge to reach us. Additionally, at the workshop I found some support for the notion that a good way to search for ETIs might be to pay more attention to the message and method we would use to initiate contact.
Comments please, by email
I was invited to present a paper at a workshop in Paris in late March 2003.
The meeting was on Encoding
Altruism: The Art and Science of Interstellar Message Composition.
A copy of my abstract can be found here.
A copy of my presentation can be found here.
In the summer of
1967 I was one of the happy band of vacation students and scientists working at
the Lords Bridge radio telescope site building the dipole array which
‘discovered’ pulsars. An account of the discovery can be found here. The dipole
array is no longer maintained – see a photo here.
Page created December 2002
Information updated on 18th February 2005
This page maintained by W.H.Edmondson
The University of Birmingham
School of Computer Science Home
Page
