Top down searches for extraterrestrial life:

1
Top down searches for extraterrestrial
life These are of two types Search for
Extraterrestrial Intelligence This is a
search for electromagnetic signals from an
extraterrestrial entity deemed by some
definition to be intelligent. It is conducted
mainly by an organization called SETI and is
mainly privately funded. Search for
extraterrestrial life This is a search
for any form of life. It is conducted
mainly by NASA using probes visiting
other planets of the solar system.
2
Electromagnetic Searches (SETI) SETI stands for
the Search for Extra Terrestrial Intelligence I
t looks in the spectrum of electromagnetic waves
coming in from space for signs of complex
intelligent life
3
Reminder of nature of electromagnetic waves
c
E
H
4
Advantages of looking for life from the
electromagnetic waves it may produce Fast they
move at the cosmic speed limit. They come to us.
Much less costly. They contain lots of
information (wide frequency band. Some of them
penetrate the interstellar medium and the
atmosphere.
5
Click on any part of the spectrum for further
detail.                                         
                                                  
                                                  
            
6
Where to look. Criteria are Waves that are
not blocked by the Atmosphere. Waves at
frequencies where there is not a lot of natural
electromagnetic noise of known origin. Carrier
frequencies which are natural ones, presumably
known to a intelligent civilization.
7
How to decide which frequencies to
explore? SETIs decision is to look where there
is not a lot of natural electromagnetic
noise. The other criterion for earth based
searches is that the radiation should be able to
get through the atmosphere. These two criteria
lead to the water hole Illustrated in the next
slide.

8
(No Transcript)
9
Origin of the 21 cm line The line marked H in
the preceding graph corresponds to
electromagnetic radiation with frequency 1430
Megahertz 1.43x109 cycles per second and wave
length ? (3x 1010 cm/s)/(1.43x109)21 cm It
is a workhorse of astrophysics. It arises
as follows Both the electron and the proton
in a hydrogen atom act like tiny magnets. The
Magnetic field of each acts on the magnetic
moment Of the other. As a result the directions
of the magnets Can precess around one another at
a characteristic frequency of 1430 MHz
10
Arecibo radio telescope 305 meter diameter.
http//www.naic.edu/public/the_telescope.htm
11
Allen telescope
Currently 42 antennas.
12
Radio searches
13
(No Transcript)
14
For more on radio searches
http//planetary.org/special/seti
http//www.seti.org/
http//setiathome.ssl.berkeley.edu/
15
The searches are of 2 types Surveys with no
special selection of stars. Usually parasitic on
investigations of Other radio astronomy
experiments (seti_at_home) Long observations of
stars selected for special Interest (Phoenix,
SETI Institute)
16
We can estimate how likely it is that the surveys
have missed a signal (at the power level of their
sensitivity and of the sort for which they are
looking.) If they have looked at M stars, then
the probability of failing to find a civilization
is P(1-Nciv/Ngal)M In the case of seti_at_home M
fNgal where f is about 1/3. If Nciv ltlt Ngal
then this can be rewritten as
Pexp(-fNciv) For f1/3, this is plotted as a
function of Nciv next.
17
(No Transcript)
18
We conclude that the probability that
the setii_at_home search is extremely unlikely to
have missed more than about 10 civilizations
emitting signals at the power level and of the
type for which they are looking.
19
But what to look for in the signals? In human
communication we use a carrier frequency.
This is what you tune the dial of a radio to.
Then we impose a signal varying more
slowly than the carrier frequency on top of the
carrier sine wave. Signal can be imposed on
the carrier sine wave by varying the amplitude
(AM or amplitude modulation) or by varying the
frequency (FM frequency modulation).
20
The existing SETI searches are looking for an
amplitude modulated signal in which the envelope
is a sine wave or a series of pulses. They dont
seem to be finding any signals of that type.
This could mean that 1. There is nothing there
at all. 2. The signals are at a lower power
level or at a different carrier frequency than
the ones being searched. 3. Not enough of the
sky has been searched. 4. They are looking for
the wrong kind of signal.
21

• I will discuss each of these a little.
• (Nothing there) is the situation if possibility
• 3 in the discussion of flife obtains. That is
• if our biosphere turns out to be a rare event.
• 2. (Power levels or carrier frequency wrong).
• The SETI searches are quite systematically
• working on this possibility, for example with
• the new Allen telescope.
• 3. (Not enough sky.) This is a possibility for
• the in depth searches (eg Phoenix) but, as
• noted, not very likely for the all sky searches
• such as seti_at_home .
• 4. (Looking for the wrong thing) is the most
  interesting
• possibility (next page).

22
4. Looking for the wrong thing? There does not
seem to be anything very compelling about looking
for a pulse or a sine wave signal except that it
can be fairly clearly distinguished from what
would be generated by most known astronomical
events. Even this is not entirely obvious
When neutron stars were discovered, the
possibility that they were extraterrestrial
civilization signals was briefly considered,
because they were radio pulses (in fact, first
called pulsars and, jocularly and informally,
LGMs (Little Green Men).
23
What are the alternatives in looking for
signals? I discuss 1. Low frequency spectral
characteristics 2. Signals like the
anthropogenic ones emitted from earth 3. Signals
which have statistical characteristics like human
language. 4. Signals which are mathematically
complex.
24
1. Low frequency characteristics. It is a poorly
understood fact that a lot of physical processes
generate a kind of noise characterized by the
fact that it gets systematically louder at lower
and lower frequencies. The general feature is
that if you graph the loudness versus the
frequency on a log-log scale you get a straight
line with negative slope
ln( noise amplitude)
ln(amplitude)-? ln (frequency)
ln(frequency f)
25
It turns out that human generate this kind of
noise spontaneously, as measured by Voss and
Clarke . They monitored radio stations of
various types in the 1970s and found the average
spectrum emittted. The results were of the above
form with ?1 (1/f noise)..
26
(No Transcript)
27
Voss and Clarke suggested that this 1/f emission
spectrum may be characteristic of life.
Therefore it might be useful to look for it in
SETI searches. Unfortunately, 1/f noise is also
generated by a lot of nonbiological sources, so
this cannot be a complete answer.
28

• Look for something like what humans
• are emitting from earth.
• Humans are not transmitting electromagnetic
• sine wave or pulse signals. We could look
• at what humans are actually emitting into space
• and seek a similar signal from other stars.
• The information about 1/f noise suggests that
• the low frequency spectrum is the most
• characteristic.
• Detailed information on the actual human
• spectrum does not seem easy to find in the
• open literature. I found the following
• Japanese data. It appears to suggest that the
• spectrum might be distinguished from the
• cosmic background.

29
(No Transcript)
30
3. Signals with statistical characteristics like
human language. There has been a lot of study
about the statistics of signaling in electrical
engineering and also in biology (songs of
humpback whales). It has not been applied to
SETI as far as I know. Some general ideas
involved As we discussed with respect to the
genome, The information I in a sequence of
symbols is I log2 (number of possible
sequences)
31
Signal continued I log2 (number of possible
messages) For example, for DNA of length N, each
monomer could have 4 values so the number of
possible messages was 4N . For English there
are, say 60 possible objects in each letter
space, giving 60N possible messages of letter
length N. But this is absolutely the MAXIMUM
number of possible messages and English is not
designed to use most of them.
32
Imagine an (untrained) monkey wacking randomly at
a keyboard. He will produce all of the 60N
messages with roughly equal likelihood and most
of them will be gibberish. By the formal
definition each of these messages will maximize
I. We could imagine redesigning English to give
a meaning to each of the 60N possible messages.
However that is not how English is currently
designed. Now one idea is that an advanced
civilization would design its language to
maximize the information content I. If it did
that, we would have no way to recognise a signal
as a signal.
33
Why there is a paper entitled A signal from a
sufficiently advanced civilization Is
indistinguishable from noise (This is by some
respectable scientists but I dont believe it is
true.) In the paper, they assume that the
advanced civilization maximises the information
content in each message. This means that they use
all the possible messages in their language with
equal frequency. Assigning them each a
meaning. But if we dont know the code, this
means that there will be no meaningful
repetitions in the message. Each letter will
look as if it has been placed randomly in the
sequence.
34
In the exercise this week you are estimating the
information content in English in a different
way, by using average word length and the number
of words in English. You will find that this
gives a smaller number for the information
content than the one which assumes that all
the gibberish collections of letters are
messages. What is different about that way of
estimating the information in an English
message? It takes what are called correlations
into account. An example of a correlation e
usually follows i
35
The letters in a message are correlated if some
letter in the message has a probability
different from 1/26 of being preceded by some
other letter. As, for example, i has a better
than random chance of preceding e.
Correlations can exist a long way back in a
message. d has a better than random chance of
being preceded by n and the set nd has a better
than random chance of being preceded by a etc.
If we just keep track of allowed words in a
language then we can see that the information
content goes down and the number of correlations
between letters goes up.
36
If you were to include the rules of grammar
in order to restrict allowed messages, then the
number of possible messages would go down even
more. The rules of grammar provide rules
concerning what words can follow other words and
thus induce further correlations in the allowed
messages, extending farther back in the
message. If you were to insist that all the
messages in a language made sense (maybe hard
to precisely define) you would restrict
the allowed messages and information even more
and extend the range of correlations
further. And so on the end of a novel is not
randomly related to its beginning, etc
37
Conclusions English does not maximize
information. It contains correlations and
repetitions which reduce the information
content. The reason for this is to make
the language easy to interpret. You can do It
with dictionary and the rules of grammar (kind of
a program) If all possible gibberish collections
of letters in English messages were allowed
you would need, for messages of length N, a code
book explaining the definitions of 26N messages
and this would be much much bigger than an
ordinary dictionary.
38
This illustrates a popular definition
of COMPLEXITY in a message The most complex
message is one which needs the largest dictionary
(or computer code) to generate or interpret it.
Thus the most complex message will carry the
most information but be the hardest to
interpret. If we judge by our own language,
languages of sophisticated biospheres will not
maximize complexity but will contain repetitions
and correlations. How do we measure the
correlations in a message without knowing the
language? For English, its been done in several
different ways. I will illustrate with one.
39
Measuring correlations in English. Arrange all
the letters in English in the order of their
frequency of use. Assign each letter a number
proportional to the frequency of its use. Thats
its probability P? . Now for each separation
n0,1,2,3,. And each pair of letters, aß find
the number times that pair appears ie Paß (n)
(number of times aß appear)/(number of pairs)
40
Now measure the correlations in messages
with C(n) SPaß(n)lnPaß(n)-2SPalnPa For
messages with maximum information, all letters
are used with the same frequency and one pair is
as likely as any other so we will find C(n) 0
for all n. In English, when this is done, one
finds something Like C(n) a 1/n4 so the
correlations extend a long way.
41
Conclusion for SETI Measure correlations in the
observed spectrum, to see if they are like those
observed in human languages. There are some
related definitions of complexity in messages
which can also be tried.