1. Thoughts
on
guerrilla
research
from
an
occasional
prac33oner.
The
brief?
“Permission
free,
no
funding,
jfdi,
quicker
to
do
research
than
write
bid..”
These
are
slides
I
didn’t
show
in
the
workshop,
but
that
I’d
sketched
as
prepara7on
trying
to
clarify
(or
maybe,
crea7vely
muddle)
my
ideas
about
what
guerrilla
research
may
or
may
not
be…
1
2. In
retrospect,
I
probably
didn’t
find
out
enough
detail
from
Mar3n
about
what
he
wanted
me
to
cover
in
the
session
–
and
I’m
s3ll
a
liHle
hazy
about
what
he
means
by
“guerrilla
research”.
So
let’s
start
with
a
bit
of
reflec3on
about
what
this
phrase
“guerrilla
research”
might
mean...
2
3. My
star3ng
point:
some
defini3ons
of
guerrilla
warfare.
These
defini3ons
all
seemed
to
agree
that
guerrilla
warfare
is
a
form
of
unconven7onal
warfare,
so
did
Mar3n
mean
unconven7onal
research?
3
4. Here’s
an
example
of
a
call
for
unconven3onal
research
from
the
US
Na3onal
Ins3tute
of
Health’s
Department
of
Health
and
Human
Services.
4
5. They’re
looking
for
novel
hypotheses,
so
possibly
things
counter
to
the
accepted
norm?
5
6. This
brief
seems
to
suggest
they’re
looking
for
research
that
maybe
goes
against
the
norm,
or
accepted
canon;
research
that
may
be
controversial,
perhaps,
in
that
it
goes
against
the
current
orthodoxy
(something
“paradigm
breaking”?)
Or
maybe
something
that
uses
a
technique
or
approach
that
maybe
hasn’t
been
tried
before,
(at
least,
not
in
the
area
of
the
call?)
6
8. Here’s
some
more
detail
from
the
same
call.
It
seems
as
if
the
DHSS
are
not
looking
for
pilot
projects.
(But
what
is
a
pilot
project
anyway?
One
designed
not
to
find
what
you’re
looking
for
but
to
show
that
you
need
to
look
further?
In
which
case,
it’s
a
real
project
designed
to
show
that
you
need
another
more
substan3al
project?)
8
9. Or
maybe
I’m
being
too
cynical…?
So
how
do
other
people
categorise
pilot
studies?
9
10. Let’s
build
the
an3cipa3on…
two
ways
–
what
could
they
possibly
be…?
10
11. Ah
ha..
A
trial
run…
which
means
a
real
run
but
not
for
real?
Something
like
a
full
dress
rehearsal
maybe?
11
12. Okay
–
this
one
makes
more
sense
to
me.
This
is
a
bit
more
like
a
technical
dress
rehearsal?
That
is,
we
take
as
long
as
we
need,
stopping
the
clock
whenever
we
come
across
a
problem
and
figure
out
how
to
solve
it?
12
14. This
presupposes
you
have
a
sort
of
idea
for
the
design
of
a
larger
study?
If
we
were
trying
to
hustle
the
funding
agency,
this
would
be
the
convincer?
14
15. So,
is
that
what
we’re
here
to
talk
about
–
doing
research
that
is
some
sort
of
pilot
project?
15
16. What’s
the
etymology
of
“guerrilla”?
Does
that
help
us
figure
out
what
guerrilla
research
may
be?
It
seems
it’s
from
the
Spanish,
meaning
li@le
war,
which
to
me
has
the
sense
of
conven3onal
war
but
on
a
small,
resistance
like
scale?
So
maybe
guerrilla
research
is
in
some
sense
li@le
research?
16
18. Small
research
grants
are
oeen
offered
by
conven3onal
funders
but
require
shorter
bid
documents
than
for
funding
larger
scale
bids.
They
may
require
less
detail
than
larger
bids,
partly
because
they
are
less
complex,
(because
less
money
means
less
things
can
be
done,
means
less
planning
and
management
is
required).
Maybe.
Where
ins3tu3ons
have
internal
stage
gate
processes
for
submigng
bids,
there
may
s3ll
be
considerable
overhead
in
pugng
even
a
small
bid
together.
18
19. Small
grants
are
typically
of
the
order
of
hundreds
to
low
thousands
of
pounds.
On
an
middle
3er
academic
salary
of
£40k,
say
£1k
per
week
(that
is
£200
per
day,
which
with
overhead
comes
in
at
double
that
-‐
£400
per
day),
if
you
spend
half
a
day
faffing
around
with
the
bid
prepara3on,
and
have
to
get
it
signed
off
internally
too,
you’ve
already
cost
your
ins3tu3on
the
minimum
amount
before
you
even
hear
whether
you’ve
got
the
grant
to
do
whatever
it
was
you
wanted.
I
know
from
wri3ng
up
blog
posts
that
describe
small
experimental
hacks
I;ve
done
that
it
can
take
over
an
hour
to
write
up/document
as
a
blog
post
a
10
minute
hack.
Working
out
how
to
describe
what
I
thought
I
wanted
to
try
to
do,
along
with
some
sort
of
‘research
ques3on’
to
jus3fy
doing
that
10
minute
hack
would
probably
take
far
longer
once
first
drae,
comments,
re-‐drae
and
sign
off
on
the
bid
are
taken
into
account.
19
20. At
first
glance
this
is
promising
–
I
can
funding
to
work
out
either
costs
of
a
very
small
project,
or
to
ask
for
project
planning
money.
But
then,
what
sort
of
project
is
legi3mate?
What
if
I
just
want
to
try
out
something
just
to
see
if
it
looks
like
it
might
be
useful
and/or
interes3ng
in
a
micro-‐blue
sky
style
approach?
How
do
I
jus3fy
that?
20
21. So
–
is
that
what
we’re
here
to
talk
about?
How
to
put
together
li@le
research
projects?
It
seems
to
me
that’s
not
what
the
phrase
evokes…
li@le
research
in
the
small
grant
sense
isn’t
unconven7onal.
21
22. Are
there
any
other
forms
of
guerrilla
ac3vity
that
might
give
us
a
steer?
[A
note
on
info
skills.
I’m
not
sure
where
or
when
I
learned
it,
but
very
early
on
in
life
I
learned
how
to
navigate
books
in
general,
and
then,
later,
how
to
navigate
technical
books,
text
books
and
academic
texts:
book
3tle
and
author
for
an
idea
of
what
was
in
the
book,
along
with
the
notes
on
the
back
cover
or
side
papers;
chapter
lis3ngs
for
gegng
an
idea
of
the
structure
of
a
book,
page
numbers
for
finding
or
remembering
specific
pages,
page
headers
for
keeping
track
or
reorien3ng
myself
within
a
book
or
naviga3ng
it
while
skimming
it,
indexes
for
finding
the
loca3on
by
page
of
a
par3cular
idea,
topic,
person,
or
place;
lists
of
figures
for
finding
out
where
the
pictures
were,
and
maybe
more
about
them;
lists
of
tables
for
finding
where
the
data
is,
and
so
on.
I
also
learned
to
navigate
the
directories
I
use
on
a
daily
basis
on
the
web:
advanced
search
pages
offer
many
ways
of
limi3ng
a
search
to
make
it
more
powerful,
but
how
many
people
use
them?
And
how
many
people
know
that
those
search
limits
are
accessible
by
addi3onal
commands
–
search
limits
–
placed
in
the
search
box.
(Indeed,
that’s
oeen
how
advanced
search
limits
are
added
to
a
search.)
In
this
case,
I
am
limi3ng
my
search
to
pages
on
the
English
version
of
Wikipedia
by
asking
only
for
results
on
h@p://en.wikipedia.org,
to
pages
that
are
actual
Wikipedia
entries
(inurl:wiki
–
look
at
the
URL/web
address
of
a
Wikipedia
page
and
you’ll
spot
why
I
added
that
par3cular
term
-‐
h@p://en.wikipedia.org/wiki/Guerrilla_gardening);
and
further
asking
that
the
word
guerrilla
appears
in
the
page
3tle.
If
you
want
to
search
content
on
the
OU
website.
A
site:open.ac.uk
limit
will
do
it
for
you.
If
you
want
to
search
across
UK
government
websites,
limit
by
site:.gov.uk;
and
so
on.
There
are
other
useful
limits
too:
filtetype:
limits
results
to
par3cular
documents
types:
filetype:ppt
for
Powerpoint
decks
(though
a
beHer
one
there
is,
in
brackets,
the
combined
(filetype:ppt
OR
filetype:pptx)
search
limit),
filetype:pdf
for
PDFs,
or
(filetype:xls
OR
filetype:xlsx
OR
filetype:csv)
to
return
spreadsheet
and
CSV
data
files.]
22
23. Guerrilla
marke7ng
is
a
form
of
marke3ng
that
subverts
the
tradi3onal
approach
to
marke3ng
in
a
couple
of
ways…
23
24. Firstly,
it
stands
counter
to
the
tradi3onal
sorts
of
“deliverable”
you
might
expect.
We’re
not
looking
for
tradi3onal
adver3sing
copy,
poster
designs
or
TV
ads…
24
25. Nor
are
we
going
to
present
the
campaign
to
the
audience
in
a
tradi3onal
way
via
tradi3onal
marke3ng
channels:
the
press,
television,
posters
or
billboards,
for
example.
25
26. So
is
that
what
we’re
here
to
talk
about?
Might
guerrilla
research
bear
any
resemblance
to
guerrilla
marke7ng,
for
example
in
the
way
it
is
a
subversion
of
tradi3onal
techniques?
26
27. Guerrilla
art
is
seen
-‐
by
Wikipedia
at
least!
–
as
a
form
of
environmental
art.
27
28. How
does
this
sound?
”The
act
of
guerrilla
research
is
focused
on
cause
and
effect,
not
the
research
piece
itself.
It
aims
to
produce
an
effect
within
the
minds
of
those
people
that
live
within
the
environment
being
altered.
It
does
not
necessarily
aim
to
produce
research
that
is
meaningful
as
research.”
[The
workshop
addi3onally
turned
up
the
idea
of
guerrilla
gardening,
sowing
seeds
or
taking
over
land
without
permission
and
pugng
it
to
use.
This
has
far
more
of
the
feel
I
think
Mar3n
was
sugges3ng…
again,
according
to
Wikipedia,
“the
act
of
gardening
on
land
that
the
gardeners
do
not
have
the
legal
rights
to
u7lize,
such
as
an
abandoned
site,
an
area
that
is
not
being
cared
for,
or
private
property.”
This
is
very
much
in
the
style
of
not
asking
for
permission,
of
iden3fying
a
valuable
but
underu3lised
or
otherwise
neglected
resource,
and
pugng
it
to
some
sort
of
use
that
is
construc3ve
within
the
environment
and
of
benefit
to
the
environment
and
its
inhabitants.]
28
29. It
seems
to
me
that
one
key
feature
of
many
guerrilla
X
interven3ons
is
that
they
are
localised,
or
at
least
represent
local
ac7ons
although
those
ac3ons
may
scale.
(For
example,
a
guerrilla
gardening
ac3vity
where
a
individuals
in
different
ci3es
or
towns
sow
a
par3cular
sort
of
seed
in
a
par3cular
sort
of
place
in
their
own
3mes
as
part
of
a
co-‐ordinated
distributed
ac3on.
Note
that
we
don’t
necessarily
need
to
localise
in
3me
either.
Segng
up
an
online
social
media
bot
to
search
for
men3ons
of
a
par3cular
brand
or
topic
and
tweet
an
autoresponse
might
be
considered
a
guerrilla
marke3ng
ac3vity.
Bots
can
be
quick
and
easy
to
set
up,
but
then
they
can
be
lee
to
free
run.
In
a
guerrilla
research
sense,
maybe
we
can
set
a
data
collec3on
ac3vity
running
(for
example,
a
tweet
stream
archiver
using
Mar3n
Hawksey’s
TAGSexplorer)
and
then
forget
about
if
for
a
month
or
two
un3l
it’s
collected
some
data
for
us?
29
31. We
formulated
some
sort
of
ques3on,
tried
to
find
resources
exploring
that
ques3on,
and
then
made
an
informal
cri3que
of
some
of
them
in
the
context
of
‘are
these
the
proper3es
we
might
ascribe
to
guerrilla
research?
31
32. Or
maybe
that
wasn’t
“research”…
We’ve
explored
“guerrilla”,
bit
not
“research”.
What
is
“research”
anyway?
32
33. Let’s
use
Wikipedia
again,
because
it’s
convenient.
We
could
use
an
academic
database,
but
Wikipedia
is
quicker
for
this
informal
study
(and
the
Wikipedia
ar3cles
are
some3mes
backed
up
by
“proper”
academic
references,
which
can
give
us
a
quick
in
(sic)
to
the
literature;-‐)
33
34. Crea7ve:
check
-‐
we
made
it
up
as
we
went
along.
Systema7c
–
ish:
a
iimited
Google
query
to
scope
the
results,
systema3c
within
the
Wikipedia
domain.
Increase
the
stock
of
human
knowledge:
we’re
trying
to
come
up
with
a
defini3on
where
one
doesn’t
already
exist.
Increase
the
use
of
this
stock
of
knowledge:
if
guerrilla
research
is
somehow
related
to
appropria3ng
resources
and
using
them
to
ask
research
like
ques3ons
where
those
ques3ons
weren’t
being
asked
(like
guerrilla
gardening
using
“spare”
ground),
then
yes,
we
are
trying
to
use
this
current
stock
of
knowledge
to
devise
new
insight,
and
possibly
new
things…
34
35. Of
course,
maybe
we
were
just
playing
at
research.
Or
maybe
we
were
just
playing…
35
36. Play
is
oeen
frowned
upon,
but
play
is
one
of
the
most
crea7ve,
and
directed
within
it’s
own
undirected/open
ended
terms,
that
I
can
think
of.
Play
is
oeen
associated
with
leisure,
or
recrea3onal
ac3vi3es,
so
if
we
were
previously
engaged
with
playing
at
research,
rather
than
guerrilla
research,
might
it
be
useful
to
explore
the
no3on
of
recrea7onal
research
and
then
see
contrast
this
with
out
emerging
sense
of
what
guerrilla
research
may
or
may
not
be?
(There
could
also
be
an
interes3ng
counterpoint
here
to
things
like
ci3zen
science,
and
amateur
science
(which
can
oeen
produce
“professional”
results,
as
for
example
in
the
case
of
“amateur”
astronomical
observa3ons.)
36
37. If
you
do
Sudoku,
or
Kakuro,
or
Killer,
or
any
other
Japanses
style
matheme3cal
puzzles,
you’re
doing
recrea7onal
mathema7cs.
There’s
actually
lots
of
it
about
–
it’s
even
a
recognised
book
category:
hHp://
www.amazon.co.uk/Recrea3onal-‐Mathema3cs-‐Science-‐Nature-‐Books/b?
ie=UTF8&node=922944
37
38. The
thing
is,
the
sort
of
problems
that
underpin
many
recrea3onal
maths
puzzles
require
you
to
use
real
maths
techniques
to
solve
them.
The
puzzle
provides
the
context
for
bringing
to
bear
a
par3cular
style
of
mathema3cal
problem
solving.
38
39. Recrea3onal
maths
also
gives
you
an
authen3c
problem
with
which
to
try
out
new
mathema3cal
ideas
or
problem
solving
techniques.
The
context
is
recrea3onal,
but
the
problem
isin
some
sense
authen3c.
But
maybe
that
was
a
diversion
–
but
it’s
worth
bearing
in
mind:
should
we
develop
the
ideas
of
both
guerrilla
research
and
recrea3onal
research.
39
40. Let’s
go
back
to
the
world
of
guerrillas.
An
area
we
might
realis3cally
class
as
research,
sort
of,
is
guerrilla
usability
tes7ng.
Usability
tes3ng
is
an
approach
used
by
designers
to
test
the
usability
of
a
design
(including
a
website
or
soeware
applica3on
user
interface
design)
with
“real”
people.
Guerrilla
usability
tes7ng
takes
this
to
the
streets.
40
41. I
first
came
across
it
via
one
of
the
most
crea3ve
people
I
know
(in
the
sense
of
follow!)
on
the
web
–
Mar3n
Belam.
41
42. Here’s
something
he
says
about
the
scien3fic
validity
of
the
approach:
“But
it
isn’t
research
science.”
It
might
be
properly
useful,
but
it’s
not
science.
So
it’s
not
proper
research?
42
43. Guerrilla
tes7ng
isn’t
just
used
by
the
crea3ve
industries
–
it’s
used
by
government
soeware
developers
too…
43
44. But
note
–
there
is
a
3ghtening
of
the
language
to
make
it
clear
what
is
and
what
isn’t
happening.
There
is
an
element
of
informality
in
the
technique…
44
46. So
–
that
was
the
prequel…
What
follows
is
what
I
started
to
talk
about…
46
47. Thoughts
on
guerrilla
research
from
an
occasional
prac33oner.
The
brief?
“Permission
free,
no
funding,
jfdi,
quicker
to
do
research
than
write
bid..”
[I
added
to
the
slides
as
Mar7n
Weller
presented,
trying
to
bring
in
addi7onal
examples
from
my
own
ed-‐tech
7nkerings
to
reflect
the
interests
of
the
#elesig
community.
Unfortunately,
I
didn’t
have
7me
to
then
prune/revise
the
narra7ve!
So
here’s
the
overkill
version!]
Some
reflec7ons
on
my
own
“prac7ce”
-‐
is
this
guerrilla
research?
47
48. I’m
going
to
split
the
talk
into
three
sec3ons,
exploring
means,
opportunity
and
mo7ve…
48
49. So
what
means
are
available
to
us
a
prototypical
guerrilla
researchers
(whatever
that
phrase
might
mean!)?
49
50. I’m
going
to
focus
on
access
to
tools,
not
just
technological
tools
and
applica3ons,
though
there
will
be
a
lot
of
those,
but
also
conceptual
and
legal
tools.
(I’ve
also
done
related
presenta3ons
on
this
under
the
theme
“Appropria3ng
IT”,
eg
hHp://blog.ouseful.info/2013/07/03/wrangling-‐data-‐with-‐free-‐tools-‐lasi13-‐workshop-‐
round-‐up/
)
50
51. Here’s
one
example:
WriteToReply.
Several
years
ago
the
UK
Government
released
a
report
called
“Digital
Britain”.
At
the
3me
I
was
interested
in
ways
of
engaging
with
government
consulta3ons
online,
so
I
posted
on
TwiHer
asking
if
anyone
had
reposted
the
original
PDF
document
in
a
blog
format
to
support
commen3ng
on
it
at
a
reasonable
level
of
granularity.
Joss
Winn,
whom
I
didn’t
know
at
the
3me,
replied,
and
within
a
couple
of
days
we’d
set
up
a
site
–
writetoreply.org
(since
shut
down)
–
and
republished
the
document
in
commentable
form.
In
the
weeks
and
months
that
followed,
we’d
set
up
a
company,
published
several
reports
by
our
own
doing
and
working
with
government
departments.
Vis
Joss,
we’d
also
got
some
JSIC
funding
to
further
develop
the
Wordpress
theme
we
were
using
as
deploy
to
a
commentable
document
playrom
for
JISC
called
JISCPress.
Read
more
about
WriteToReply
here:
hHp://blog.ouseful.info/?
s=writetoreply&order=asc
51
52. As
someone
who
plays
with
tech
a
lot,
I’ve
no3ced
how
it’s
got
much
easier
to
do
some
things
over
3mes
(as
well
as
harder
to
others
as
various
services
shut
down
features
that
make
them
“hackable”
in
the
sense
of
easily
appropriated).
52
53. Maps
are
a
good
example
of
this…
Several
years
ago
I
had
on
my
mental
to-‐do
list
“learn
how
to
put
markers
on
maps”.
I
was
wai3ng
for
an
appropriate
data
set
to
turn
up,
and
one
did:
the
newly
launched
Guardian
data
blog
published
a
spreadsheet
of
MPs
travel
and
office
expenses.
(This
was
before
the
MPs’
expenses
scandal
arising
from
the
release
of
individual
receipts
–
totalled
expenses
by
spending
area
per
MP
had
been
released
on
an
annual
basis
for
years;
the
Guardian
just
made
it
easier
to
work
with
by
publishing
it
via
a
Google
spreadsheet).
I
played
with
the
data,
producing
a
range
of
“tradi3onal”
visualisa3ons
–
histograms
to
count
the
number
of
MPs
claiming
a
par3cular
amount
in
a
par3cular
spending
area,
scaHerplots
to
look
for
(an3)correla3ons
between
spending
areas
(office
expenses
and
postage,
for
example,
or
rail
fares
vs
air
fares).
I
also
learned
how
to
put
markers
on
maps
–
colouring
markers
according
to
the
value
of
a
claim
in
a
par3cular,
selected
spending
area
and
markers
placed
on
the
mid-‐point
of
the
MPs
cons3tuency.
In
most
cases,
MPs
claimed
similar
amounts
to
theit
neighbours,
but
in
others
they
were
out
of
kilter;
the
map
helped
iden3fy
such
differences.
(Another
way
of
revealing
such
informa3on
might
be
to
plot
expense
type
vs
distance
or
travel
3me
from
the
cons3tuency
to
Westminster).
53
54. As
well
as
learning
how
to
plot
maps,
I
learned
how
to
draw
boxes,
in
this
case
represen3ng
the
bounding
box
around
cons3tuencies
to
see
if
area
might
be
related
to
the
size
of
a
par3cular
expense
type.
At
the
3me
it
was
hard
to
plot
maps
that
displayed
cons3tuency
boundaries,
colouring
in
consituencies
by
the
size
of
claim
to
produce
choropleth
maps.
The
release
of
shapefiles
as
open
data
on
the
one
hand,
and
development
of
free
online
mapping
applica3ons
on
the
other,
has
made
this
much
easier
to
do
nowadays.
54
55. Having
learned
how
to
draw
boxes,
I
also
learned
how
to
draw
circles,
again
using
the
MP’s
travel
expenses
as
a
foil.
In
this
case,
the
area(?)
is
related
to
the
size
of
a
par3cular
expense
type
and
the
colour
is
by
party:
did
claim
sizes
appear
to
follow
party
poli3cal
lines?!
55
56. An
idea
that
par3cular
intrigued
as
a
possible
example
of
sort-‐of-‐academic
research
posted
non-‐tradi3onally
can
be
found
pn
the
Prochronism
blog.
(A
prochromism
is
a
par3cular
sort
of
anachronism,
where
a
word
or
phrase
is
used
in
story
set
in
a
par3cular
period,
for
example,
earlier
than
the
phrase
appears
in
commonly
reported
language.)
56
57. The
original
prochromism
blog
post
(which
originally
appeared
on
the
author’s
personal
blog
before
he
set
up
the
Prochronism
blog
to
collect
together
these
associated
posts)
describes
how
tradi3onal
approached
to
anachronism
research
might
proceed.
57
62. And
the
technique
can
be
employed
at
the
speed
of
life,
rather
than
the
speed
of
most
academic
research
proposals…
62
63. If
guerrilla
research
is
a
strike
against
the
conven3onal,
it
may
provide
a
context
for
exploring
novel
coding
schemes,
either
developed
de
novo,
or
in
the
context
of
impor3ng
a
technique
common
in
one
discipline
into
another
where
it
is
not
used.
63
64. A
technique
I
have
found
useful
I
first
saw
used
by
MaH
Morrison
(@mediaczar)
[hHp://blog.magicbeanlab.com/networkanalysis/how-‐should-‐page-‐admins-‐deal-‐with-‐
flame-‐wars/
].
We
had
both
been
learning
about
genera3ng
charts
using
the
ggplot2
library
in
R,
and
swapping
techniques
we
had
learned.
One
chart
in
par3cular
jumped
out
at
me,
not
least
because
the
coding
schemed
it
use
was
so
simple,
yet
it
produced
some
startlingly
original
charts
(to
me
at
least).
The
chart
type
is
a
scaHerplot;
along
the
x-‐axis
we
have
a
3me
base,
in
this
case,
the
‘number’
of
a
post
on
a
Facebook
wall.
On
the
y-‐axis,
we
have
accession
number
of
individusal
pos3ng
on
to
the
wall.
The
first
individual
has
accession
number
1,
the
second
accession
number
2,
and
so
on.
If
someone
returns
to
post
several
3mes,
we
use
the
accession
number
from
the
first
3me
we
saw
them.
This
technique
–
which
we
started
to
call
accession
plots,
or
accession
charts
–
was
completely
new
to
me.
And
very
generalisable.
64
65. Here’s
an
example
of
an
accession
chart
I
created
around
a
TwiHer
hashtag.
Aeer
collec3ng
tweets
that
contained
the
tag,
I
ploHed
them
using
the
tweet
crea3on
3mestamp
on
the
x-‐axis.
On
the
y-‐axis
I
ploHed
TwiHer
screen-‐names,
ordering
the
names
according
to
the
order
in
which
users
first
used
the
hashtag
(that
is,
their
accession
to
the
hashtag
usage).
Ver3cal
lines
to
the
lee
show
that
a
large
number
of
people
(rela3vely
speaking)
use
the
hashtag
for
the
first
3me
over
a
short
period
of
3me;
a
large
number
of
dots
along
a
horizontal
line
show
a
user
is
par3cularly
prolific
in
their
use
of
the
hashtag.
65
66. Here’s
the
same
chart
as
before,
with
an
addi3onal
informa3on
layer:
tweets
are
coloured
as
to
whether
they
are
a
retweet
or
a
new
tweet.
This
way
we
can
see
whether
nor
not
we
have
a
retweet
burst,
or
maybe
a
conversa3on…?
66
67. Here’s
another
TwiHer
chart,
again
using
an
accession
number
device
on
the
y-‐axis,
but
this
3me
related
to
the
accession
number
of
followers
of
an
individual
[hHp://blog.ouseful.info/2013/04/05/es3mated-‐follower-‐
accession-‐charts-‐for-‐twiHer/
].
(If
you
get
the
friends
or
followers
list
of
someone
on
TwiHer,
it
is
in
reverse
chronological
order.)
In
this
chart,
accession
number
1
is
the
first
person
to
follow
the
named
individual,
number
2
the
send
person
to
follow
them,
and
so
on.
The
x-‐axis
the
number
of
days
ago
(from
the
3me
the
chart
was
generated)
that
each
follower
had
first
joined
TwiHer.
The
chart
thus
plots
accession
number
when
following
a
specified
individual
against
3me
since
joining
twiHer
(in
days).
We
see
two
features
in
the
chart:
a) a
sharp
edge
1500
days
ago,
which
corresponds
to
a
3me
when
the
number
of
TwiHer
users
in
general
exploded;
b) A
cut
off
line,
marked
red,
that
provides
an
es3mate
of
the
date
when
follower
with
accession
number
N
started
following
the
target
individual.
Generally,
this
informa3on
is
not
available
–
the
follower
list
orders
the
followers
of
an
individual
but
doesn’t
tell
you
when
they
started
following.
However,
note
that
person
X
cannot
follow
person
Y
before
person
X
joins
TwiHer.
As
accession
number
y-‐increases,
if
we
keep
track
of
the
most
recent
TwiHer
user
crea3on
date
seen
so
far
(the
right
most
point
seen
to
date)
and
plot
that
in
red,
we
get
an
es3mate
of
when
users
started
to
follow
the
target.
(Read
it
this
way:
suppose
that
in
week
M,
a
user
joins
TwiHer
and
immediately
follows
the
target
account
on
date
dM,
gaining
follower
accession
number
aM
for
that
account,
user
with
accession
number
aM+1
can’t
have
started
following
the
target
un3l
at
least
date
dM,
even
if
both
they
and
the
target
account
have
been
on
TwiHer
for
many
months
prior
to
that
date.)
The
line
chart
at
the
boHom
of
the
graph
is
actually
derived
data
that
provides
a
count
of
how
many
people
are
es3mated
to
have
started
following
the
target
on
each
day.
In
this
case
we
see
a
spike
440
or
so
days
ago.
This
chart
actually
corresponds
to
an
MP
–
the
day
they
got
a
sharp
increase
in
followers
was
the
day
they
were
elected.
Looking
up
the
dates
corresponding
to
spikes
on
other
MPs’
follower
accession
chart
in
news
archives
turns
up
other
similar
effects,
as
well
as
scandal
stories
that
hit
the
news,
were
shared
on
TwiHer,
and
lead
to
people
following
the
MP
as
a
result
[hHp://blog.ouseful.info/2013/03/04/what-‐happened-‐then-‐using-‐
approximated-‐twiHer-‐follower-‐accession-‐to-‐iden3fy-‐poli3cal-‐events/
].
Having
shared
this
technique
via
my
blog,
several
other
people
picked
it
up
and
started
using
it
in
more
formal
research
[hHp://mappingonlinepublics.net/2013/07/08/introducing-‐twiHer-‐follower-‐accession-‐graphs/
].
67
68. I
was
also
contacted
by
a
UK
journalist
to
inves3gate
whether
one
par3cular
MP
had
been
buying
followers.
I
generated
the
follower
accession
chart
and
came
to
the
conclusion
they
had
been
aHacked
by
spam
bots…
The
sudden
growth
in
followers
is
due
to
large
numbers
of
followers
with
batched
crea3on
dates
(i.e.
machine
generated)
signing
up
as
followers
in
a
very
short
period
of
3me.
Bought
followers
can
oeen
being
machine
generated
and
maintained,
but
you’d
have
to
be
really
cheap
to
buy
such
obvious
ones…
Note
that
this
signature
more
the
machine
generated
accounts
is
easy
to
spot:
but
apparently
not
so
easy
that
TwiHer
can
spot
them
and
block
them
automa3cally…
68
69. We’re
all
familiar
with
the
idea
of
using
a
microscope
to
look
at
the
very
small.
By
a
similar
token,
macroscopes
allow
us
to
look
at
everything
within
a
dataset
(“N=all”).
69
70. For
some
years
the
OU
has
been
publishing
open
data
(on
data.open.ac.uk)
about
OU
courses
and
resources.
One
of
the
datasets
lists
courses
and
courses
they
are
related
to.
Grabbing
a
copy
of
this
whole
dataset,
then
graphing
connec3ons
between
courses
that
are
related
to
each
other
and
mapping
the
result
using
a
force
directed
network
layout
algorithm
that
tries
to
posi3on
nodes
that
are
connected
to
each
other
close
to
each
other,
we
can
generate
a
map
that
shows
how
OU
courses
relate
to/cluster
with
each
other
[hHp://blog.ouseful.info/2011/01/30/open-‐university-‐
undergraduate-‐module-‐map/
].
Try
gegng
such
a
macroscopic
view
from
the
OU
courses
website…
70
71. This
is
a
macroscopic
view
over
MP
vo3ng
behaviour
over
a
parliament
several
governments
ago
(data
was
grabbed
from
the
public
whip
website,
I
think?).
Each
row
is
an
MP,
the
rows
grouped
by
party
(Labour,
the
government
at
the
3me,
is
the
top
block;
then
LibDems,
then
Conserva3ves,
then
Other).
Each
column
is
a
separate
division/vote
in
the
House
of
Commons.
The
colour
show
whether
the
MP
voted
for
or
against
the
mo3on
(I
think?!)
[
hHp://blog.ouseful.info/2010/04/22/visualising-‐
whether-‐the-‐libdems-‐side-‐with-‐the-‐tories-‐or-‐labour-‐in-‐parliamentary-‐votes/
].
Once
you
get
your
eye
in,
you
see
that
the
LibDems
tended
to
vote
with
the
Conserva3ves
in
many
case.
When
you
really
get
your
eye
in,
you
can
also
spot
rebels.
The
black
horizontal
lines
are
where
an
MP
didn’t
vote
–
possibly
because
they’re
a
minister
doing
other
things…
(This
was
actually
an
interac3ve
visualisa3on
generated
using
Processing
–
you
could
hove
over
points
to
find
the
name
of
each
MP,
the
par3cular
vote,
etc.)
The
idea
of
this
visualisa3on
nd
the
summaries
and
analy3c
ques3ons
is
suggests
is
part
of
the
value
of
this
piece,
rather
than
it’s
u3lity
as
a
visualisa3on
of
the
data
itself.
Here
are
some
other
experiments
using
another
source
of
vote
data,
this
3me
from
general
elec3ons:
hHp://blog.ouseful.info/2010/05/03/playing-‐with-‐processing-‐arc-‐
and-‐general-‐elec3on-‐data-‐2005/
71
72. Every
presenta3on
I
do,
I
try
to
get
some
Formula
One
data
in!
This
is
data
grabbed
from
the
McLaren
live
dashboard,
an
online
interac3ve
that
McLaren
ran
for
several
years
that
streamed
telemetry
data
rela3ng
to
speed,
“g-‐
force”,
throHle
and
brake
control,
gear,
distance
round
circuit
and
la3tude
and
longitude
of
the
two
McLaren
cars
during
race
weekends
[
hHp://blog.ouseful.info/
2010/04/07/f1-‐data-‐junkie-‐driver-‐dna/
].
The
line
charts
on
the
lee
are
a
typical
display.
The
right
charts
I
called
DNA
charts
–
distance
round
the
circuit
is
on
the
horizontal
x-‐axis,
lap
number
on
the
y-‐axis.
The
charts
show
the
remarkable
consistency
of
the
drivers.
The
top,
blue
strip
shows
the
gear
(1
to
7);
the
green
strip
shows
the
throHle
pedal
depression
(0-‐100%),
and
the
red
strip
shows
the
brake
(0-‐100%).
The
light
blue
strip
is
a
composite
of
the
previous
three
strips.
The
whiter
the
pixel,
the
closer
it
is
to
100%
throHle
in
7th
gear
with
no
braking.
The
boHom
two
traces
show
the
longitudinal
and
lateral
g-‐force
respec3vely.
For
the
longitudinal
trace,
red
shows
braking
–
being
forced
into
the
steering
wheel;
green
shows
accelera3on
–
being
forced
back
into
your
seat.
You’ll
see
the
greatest
g-‐force
under
braking
occurs
when
the
brakes
are
slapped
full
on…
(the
red
bits
in
the
third
and
fieh
traces
line
up).
For
the
la3tudinal
g-‐force,
the
red
shows
the
driving
being
flung
to
the
lee
(i.e.
right
hand
corner),
the
green
shows
them
being
pushed
out
to
the
right.
72
73. We
can
also
pair
the
DNA
charts
of
the
two
McLaren
drivers,
and
then
look
for
differences…
[
hHp://blog.ouseful.info/2010/04/18/f1-‐data-‐junkie-‐mclaren-‐driver-‐
comparison-‐snapshots/
]
Midway
round
the
circuit,
we
no3ce
the
NGear
traces
markedly
differ,
for
example.
73
74. Here’s
one
example
of
where
the
traces
differ
at
a
par3cular
point
round
the
circuit.
74
75. We
can
rebase
the
chart
to
use
a
2d
plot
loca3ng
the
points
according
to
la3tude
and
longitude
values,
rather
than
distance
round
the
track
to
see
where
on
a
more
tradi3onal
circuit
layout
the
differences
occur.
(I
have
also
offset
the
two
drivers
traces
so
we
can
see
them
beHer
–
they
are
not
taking
such
radically
different
lines!)
75
76. Another
important
element
of
Means
relates
to
data,
and
in
par3cular
data
sources
that
we
can
reuse
for
our
own
analyses.
Collec3ng
data
is
oeen
a
major
part
of
research
exercises,
but
in
guerrilla
research
maybe
we
should
focus
more
on
appropria3ng
and
reusing
data
that
already
exists.
If
I
only
have
half-‐an-‐hour
to
do
something
interes3ng,
I
can’t
spend
six
months
collec3ng
data…
But
I
may
be
able
to
download
something
relevant
in
seconds…
Maybe
we
don’t
need
more
data
–
maybe
we
just
need
to
spend
a
liHle
more
3me
looking
at
how
we
can
piece
together
data
that
already
exists
and
ask
ques3ons
across
it
in
a
form
of
what
we
might
term
combinatorial
data
analysis.
(If
I
have
three
data
sets,
A,
B,
C
that
share
a
common
column
that
allows
them
to
be
combined,
I
can
analyse:
just
A,
just
B,
just
A,
A
and
B
combined,
A
and
C
combined,
B
and
C
combined,
A
and
B
and
C
combined.
Most
people
will
have
researched
on
A
or
B
or
C.
In
a
typical
research
project
I
might
then
collect
D.
Maybe
we
should
start
looking
to
see
if
we
can
analyse
the
combina3ons
instead?
(Of
course,
there
are
many
reasons
why
combina3ons,
even
if
possible,
may
not
be
valid.
But
combinatorics
suggests
there
are
a
large
number
of
possible
combina3ons
that
may
be
valid
as
we
increase
the
number
of
combinable
datasets
available)).
76
77. In
the
UK,
recent
years
has
seen
cross-‐party
support
for
the
release
of
public
data
under
an
open
license
that
allows
it
to
be
shared
and
reused.
Data.gov.uk
is
a
catalogue
that
covers
data
releases
from
across
UK
government
and
other
public
services.
But
is
there
a
land
grab
going
on?
All
data
is
poli3cal,
and
it
seems
that
cataloguing
it
is
poli3cal
too.
As
the
Government
Digital
Service
(GDS)
takes
over
the
website
opera3ons
of
more
and
government
departments
within
its
gov.uk
domain,
it
can
oeen
be
more
convenient,
and
more
complete,
to
search
departmental
content
published
on
gov.uk
than
data.gov.uk
–
the
laHer
requires
human
effort
to
add
catalogue
records
to
point
to
content
that
has
been
published
on
gov.uk,
whereas
once
on
gov.uk,
it
can
be
discovered
more
directly.
77
78. As
well
as
open
data
published
by
the
public
sector,
academic
research
is
star3ng
to
be
opened
up
too.
78
79. If
public
money
has
funded
the
produc3on
of
(research)
data,
that
data
should
be
available
to
the
public,
or
so
the
argument
goes…
79
80. Some
journals
too
are
making
it
a
requirement
that
data
is
published
alongside
research
papers,
not
least
so
the
analyses
that
appear
in
those
papers
can
be
replicated
using
the
same
data.
80
81. As
well
as
data
published
openly
and
either
freely
or
at
cost,
we
can
also
request
data
using
Freedom
of
Informa3on
legisla3on
(as
well
as
the
Data
Protec3on
Act
for
data
about
ourselves,
and
data
covered
by
environmental
protec3on
regula3ons).
81
82. Note
that
despite
the
driver
from
the
Research
Councils
UK
that
more
academic
research
data
is
openly
shared,
and
despite
the
fact
that
publicly
funded
university
research
is
FOIable,
there
are
exemp3ons
from
releasing
research
data
under
FOI…
82
83. If
you
haven’t
made
an
FOI
request
before,
and
you’re
happy
for
it
to
be
made
publicly,
whatdotheyknow.com
makes
it
easy:
select
the
pubic
organisa3on
you’d
like
to
make
a
request
to,
and
you
can
send
an
email
directly
to
the
right
address.
Any
responses
are
managed
by
the
service.
If
you
browse
through
responses
to
requests,
you
see
many
of
them
include
data
files
(CSV
files
or
Excel
spreadsheets).
A
quick
hack
I
produced
[hHp://blog.ouseful.info/
2012/04/28/the-‐foi-‐route-‐to-‐real-‐fake-‐open-‐data-‐via-‐whatdotheyknow/
]
indexed
the
requests
that
returned
data
files
so
I
could
use
it
as
an
index
of
FOId
data.
(Note
that
just
because
data
is
released
under
FOI
it
doesn’t
mean
it’s
openly
licensed…)
Not
all
FOI
requests
are
made
through
WhatDoTheyKnow,
of
course
(journalists
wouldn’t
take
to
make
requests
made
as
part
of
an
inves3ga3on
available
via
a
public
service
where
other
people
can
see
what
they
are
reques3ng).
Informa3on
about
FOI
requests
made
to
organisa3ons
is,
however,
public
informa3on…
Some
organisa3ons
rou3nely
publish
a
disclosure
log,
where
they
publish
informa3on
about
requests
and
responses
with
personal
informa3on
removed.
In
other
cases,
you
may
have
to
FOI
the
same
informa3on…
83
84. The
Guardian
Data
Store
has
been
republishing
public
data
via
Google
Spreadsheets
for
some
3me.
Each
year,
it
publishes
the
data
used
for
its
university
rankings
tables.
This
example
[hHp://blog.ouseful.info/2012/09/04/filtering-‐guardian-‐university-‐data-‐
every-‐which-‐way-‐you-‐can/
]
shows
how
I
used
the
Google
Visualisa3on
API
to
provide
a
quick
tool
for
exploring
the
rankings
based
on
selec3vely
filtering
across
each
of
the
ranking
factors.
This
year,
I
used
the
R
Shiny
library
to
produce
an
interac3ve
explorer
using
R:
hHp://
blog.ouseful.info/2013/06/21/disposable-‐visual-‐data-‐explorers-‐with-‐shiny-‐guardian-‐
university-‐tables-‐2014/
84
85. If
you
don’t
feel
comfortable
building
your
own
applica3on
from
lines
of
code
(even
if
it
only
takes
10
or
20
lines
of
code
you
can
largely
copy
and
paste
from
other
people
who’ve
done
similar
things
before…)
tools
like
Google
Fusion
Tables
allow
you
to
interac3vely
explore
quite
large
datasets.
The
example
shown
here
provides
an
environment
for
exploring
chari3es
data
[
hHp://blog.ouseful.info/2013/05/01/a-‐
quick-‐peek-‐at-‐some-‐chari3es-‐data/
].
Whilst
Fusion
Tables
look
like
spreadsheets,
they
have
several
benefits:
1) they
can
be
used
to
store
much
larger
datasets
than
you
can
load
in
to
a
spreadsheet;
2) it’s
easy
to
merge
different
tables
that
share
a
common
column
(hence
“fusion”
tables?).
If
VLOOKUP
confuses
you,
this
makes
it
much
easier
and
works
across
tables
too;
3) you
can
add
filters
to
tables
to
see
just
the
informa3on
you
want;
4) genera3ng
pivot
table
style
summary
reports
is
easy
(and
these
work
across
filtered
data
too);
5) genera3ng
charts
is
easy
(and
these
work
across
filtered
data
too);
6) If
you
address
data,
Google
Fusion
Tables
can
geocode
it
for
you
too,
so
you
can
add
markers
to
a
map,
and
colour
them
by
data
values;
7)
if
you
have
shapefile
data
or
data
that
can
be
merged
with
shapefiles
(eg
MP
cons3tuencies),
you
can
use
Google
Fusion
Tables
to
make
choropleth
maps.
85
86. When
is
a
thing
the
same
as
another
thing?
Is
Poppleton
University
the
same
as
the
University
of
Poppleton?
Is
the
laHer
the
same
as
the
University
of
Poppelton?
What
would
a
search
for
“Poppleton”
turn
up?
86
87. OpenRefine
(from
openrefine.org)
is
a
cross-‐playorm
browser
based
applica3on
for
cleaning
and
reshaping
datasets.
It
has
something
of
the
look
of
a
spreadsheet
applica3on
about
it,
in
that
it
works
with
tabular
data,
but
it
has
been
designed
for
gegng
your
data
into
a
state
and
a
shape
where
you
can
start
to
work
with
it.
(As
well
as
opening
spreadsheet
files,
CSV
files,
a
wide
range
of
text/line
item
based
data
files,
it
can
open
XML
files
and
JSON
files
and
help
you
get
them
into
a
tabular
format.)
One
of
the
tools
it
offers
is
to
“cluster”
similar
elements
appearing
within
a
data
column.
There
are
several
well-‐known
algorithms
for
trying
to
do
this
that
OpenRefine
supports.
Running
a
clustering
algorithm
iden3fies
items
that
are
different-‐but-‐might-‐actually-‐be-‐the-‐same,
and
gives
you
the
op3on
of
rewri3ng
them
automa3cally
so
they
are
the
same.
It
beats
working
through
the
files
by
hand…
87
88. OpenRefine
also
gives
you
a
way
in
to
the
world
of
Linked
Data
and
the
seman3c
web.
OpenRefine
can
look
up
items
within
a
column
against
Linked
Data
sources
and
retrieve
canonical
iden3fiers
for
them.
88
89. These
iden3fiers
can
then
be
used
to
pull
back
data
associated
with
(that
is,
“linked”
to)
those
items…
89
90. So
the
means
of
discovering
and
obtaining
data
that
already
exists,
as
well
as
finding
tools
that
can
work
wonders
with
that
data,
are
increasingly
out
there.
But
do
we
have
any
opportunity
to
make
use
of
those
resources
without
the
backing
of
a
formal
–
and
funded
–
research
project?
The
guerrilla
research
mentality
of
“just
doing
it”
suggests
we
could
use
the
3me
that
would
otherwise
be
spent
wri3ng
bids
actually
doing
the
(guerrilla)
research
just
anyway…
but
then,
at
some
point
we
have
to
become
accountable
(or
maybe
we
don’t!).
So
let’s
consider
what
opportuni3es
there
are
for
doing
guerrilla
research
that
we
may
be
able
to
jus3fy
by
other
means
if
called
to
account…
90
91. Several
years
ago
I
looked
aeer
an
OU
short
course
that
was
delivered
largely
online
but
with
some
offline
reading
and
ac3ves.
At
the
3me
I
was
interested
in
the
extent
to
which
we
could
use
web
analy3cs
to
analyse
the
performance
of
a
course
delivery
website
as
a
website,
something
that
s3ll
doesn’t
really
feature,
as
far
as
I
can
tell
(learning
analy7cs
tend
to
focus
on
slghtly
different
concerns,
and
has
the
poten3al
to
be
far
more
misleading
and
malevolent…)
The
above
chart
shows
the
course
pages
segmented
into
groups,
with
each
group
containing
the
pages
related
to
a
par3cular
week’s
ac3vi3es.
Time
is
along
the
horizontal
x-‐axis,
some
measure
of
ac3vity
on
the
ver3cal
y-‐axis.
The
chart
shows
that
the
students
appear
to
work
through
the
course
as
paced,
returning
to
the
content
of
earlier
weeks
as
the
end-‐of-‐course
assessment
deadline
looms.
Web
analy3cs
have
come
on
some
way
since
then,
and
I’d
track
and
analyse
things
slightly
differently
now;
but
I
think
there’s
s3ll
a
lot
that
can
be
done
in
terms
of
understanding
how
online
courses
work
as
websites
that
can
feed
back
into
the
course
design.
(A
no-‐brainer
is
tracking
which
links
are
clicked
on;
if
no-‐one
ever
clicks
on
a
par3cular
resource
link,
what
use
is
it?
If
it’s
key,
you
need
to
find
new
ways
of
encouraging
students
to
click
it…)
There
are
of
course
ethical
and
privacy
issues
associated
with
using
Google
Analy3cs
–
you
tell
Google
every
page
that
each
of
your
students
has
visited
on
the
site,
and
when.
And
Google
could
in
principle
generate
a
marke3ng
group
based
on
your
cohort
from
the
set
of
individuals
accessing
that
set
of
pages.
(For
related
considera3ons,
see:
hHp://
blog.ouseful.info/2010/05/17/personal-‐declara3ons-‐on-‐your-‐behalf-‐why-‐visi3ng-‐one-‐
website-‐might-‐tell-‐another-‐you-‐were-‐there/
)
Old
presenta3on:
hHp://www.slideshare.net/psychemedia/course-‐analy3cs-‐in-‐context-‐
presenta3on
91
92. As
well
as
dabbling
with
googaly3cs
on
course
webpages,
something
I
wasn’t
supposed
to
do,
I
also
managed
to
get
access
(with
permission)
to
the
webstats
for
the
OU
Library.
Again,
I
was
interested
in
seeing
what
we
could
learn
about
how
well
the
site
was
working
as
a
website.
And
then
I
posted
some
thoughts
and
learnings
about
it…
92
93. For
a
long
term,
search
was
my
passion,
wondering
how
we
could
appropriate
search
technologies
for
our
own
ends.
We’ve
already
seen
how
search
limits
can
be
used
to
refine
a
web
search
so
that
results
can
be
limited
to
results
of
a
par3cular
sort
(from
a
par3cular
domain,
of
a
par3cular
document
type,
or
containing
a
par3cular
word
in
the
3tle,
for
example).
One
tool
for
industrialising
this
is
a
custom
search
engine
such
as
a
Google
Custom
Search
Engine.
These
search
engines
can
be
configured
to
return
sets
from
a
par3cular
set
of
web
pages
or
domains.
To
a
certain
limited
extent
you
can
also
tune
the
rankings.
Over
several
years,
I
dabbled
with
ways
of
dynamically
selec3ng
the
resources
that
custom
search
engines
would
search
over.
But
always
at
the
back
of
my
mind
was
whether
a
course
custom
search
engine
would
be
useful.
That
is,
for
courses
that
have
lots
of
links
to
web
pages
or
other
online
resources,
could
we
make
a
useful
search
engine
based
around
those
resources?
For
example,
could
we
extract
the
links
contained
in
the
course
materials
for
a
par3cular
course
(yes
we
can,
it’s
easy)
and
use
these
as
the
basis
of
a
custom
search
engine,
i.e.
one
that
would
search
over
the
resources
listed
in
the
course,
and
other
poten3ally
rela3ve
content
(or
content
of
a
reputable
quality,
by
associa3on)
from
the
domains
the
linked
to
content
was
published
on.
Well,
yes,
we
can
do
that,
but
from
my
dabblings,
it’s
a
bit
rubbish
–
the
course
custom
search
engine
doesn’t
have
a
big
enough
index
to
be
useful
as
a
search
engine,
even
within
a
limited
domain.
More:
hHp://blog.ouseful.info/2011/11/08/notes-‐on-‐custom-‐course-‐search-‐engines-‐
derived-‐from-‐ou-‐structured-‐authoring-‐documents/
93
94. One
of
the
many
great
things
about
the
OU
is
the
way
the
ins3tu3on
has
engaged
with
the
publishing
of
open
content
in
standardised
formats.
The
course
material
web
pages
published
on
the
OpenLearn
website
are
rendered
(as
are
OU
‘actual’
course
materials)
from
a
structured
XML
document
format.
I’m
not
sure
if
you
s3ll
can,
but
you
certainly
used
to
be
able
to
get
hold
of
the
underlying
XML
document
that
provided
the
‘source
code’
for
OpenLearn
course
materials
just
by
hacking
around
with
the
URL.
So..
Play3me,
right?
In
one
experiment,
I
tried
genera3ng
interac3ve
mindmaps
as
alterna3ve
naviga3on
surfaces
over
OpenLearn
materials
(hHp://blog.ouseful.info/2012/05/04/
genera3ng-‐openlearn-‐naviga3on-‐mindmaps-‐automagically/
).
This
harked
back
to
a
more
bespoke
approach
I’d
used
in
a
previous
OU
course
where
I’d
created
a
mindmap
by
hand
to
provide
students
with
an
alterna3ve
way
of
naviga3ng
the
online
course
materials.
One
advantage
of
genera3ng
mindmaps
automa3cally
was
that
I
could
put
in
a
generic
search
term
and
generate
a
mindmap
style
way
of
naviga3ng
over
all
OpenLearn
resources
that
reference
the
par3cular
search
term.
Another
quick
hack,
as
depicted
in
the
slide,
was
to
create
a
gallery
of
all
the
images
contained
in
OpenLearn
course
resources,
and
provide
a
search
over
them.
It
only
took
changes
to
a
couple
of
lines
of
code
to
then
produce
a
search
tool
that
covered
glossary
items
from
across
the
OpenLearn
course
content.
These
quick
tools
could
easily
be
hacked
up
around
all
OU
course
materials
for
use
internally
as
ad
hoc
tools
to
help
support
course
development,
for
example.
But
they
haven’t
been.
If
a
project
isn’t
big
enough
to
aHract
a
budget
code
and
kudos
for
a
manager,
it
won’t
be
pursued.
Guerrilla
projects
are
irresponsible,
and
without
responsibility,
they
won’t
be
adopted…
At
least,
not
formally
;-‐)
94
95. I
love
the
word
“finesse”,
as
for
example
in
cards
or
chess
where
you
get
something
extra,
for
free,
that
maybe
you
shouldn’t
have
expected
to
have
a
right
to.
By
“finessing
permission”,
I
mean
something
complementary
to
asking
for
permission
qua
forgiveness,
aeer
the
fact…
Rather,
I
mean
something
more
akin
to
retrofi_ng
permission,
finding
some
ra3onale,
maybe
even
a
post
hoc
ra3onale*,
that
let’s
you
jus3fy
a
guerrilla
research
ac3on.
*
As
anyone
who
has
ever
wriHen
up
a
piece
of
formal
research,
the
way
it
happened
is
not
the
way
you
write
it
up.
Another
reason
why
formal
research
reports
are
oeen
rubbish
when
it
comes
to
helping
other
figure
out:
a)
what
you
did,
b)
why
you
did
it,
and
c)
how
you
figured
out
how
to
do
it
that
way.
Blog
posts
as
a
research
notebook
posts
are
far
more
authen3c,
and
far
more
useful
for
helping
people
figure
out
your
method
and
methods
of
their
own.
95
96. As
well
as
course
materials
extracted
from
“official”,
for
credit
OU
courses,
OpenLearn
also
published
material
to
support
the
various
broadcast
offerings
that
the
OU
co-‐
produces
with
the
BBC.
Every
so
oeen
I
submit
an
ar3cle
to
the
OpenLearn
editorial
team,
or
respond
to
a
request
from
them,
to
wrap
a
feature
on
the
the
Radio
4
programme
More
or
Less
or
to
pick
up
on
a
current
news
story.
96
97. Time
is
money,
supposedly.
So
if
you
need
to
get
money
into
the
equa3on,
or
at
least,
the
promise
of
it,
(how
much
3me
is
spent
preparing
bids
in
the
hope
that
one
of
them
pays
off?!),
what
op3ons
are
there?
97
98. If
you
know
the
story
of
“Longitude”,
you
know
about
prize
funds.
Rather
than
gegng
people
to
compete
for
money
based
on
things
they
say
they
are
going
to
do,
government
or
government
agencies
set
up
a
challenge
and
then
reward
the
best
entry;
or
they
don’t,
if
the
best
entry
isn’t
good
enough.
The
thing
that’s
now
Google’s
autonomous
car?
DARPA
bootstrapped
that
with
their
autonomous
vehicle
Grand
Challenge.
The
EU
is
looking
to
use
inducement
prices
as
part
of
its
funding
strategy.
And
the
research
councils
keep
experimen3ng
too:
the
currently
open
“Visualising
Research”
compe33on
encourages
people
to
submit
visualisa3ons
and
visualisa3on
applica3ons
around
data
about
UK
research
awards,
as
published
via
the
Gateway
to
Research.
So
if
you
want
to
learn
how
to
get
JSON
data
out
of
an
API,
or
want
to
learn
how
to
create
a
par3cular
sort
of
visualisa3on,
steal
some
3me
and
have
a
play
with
the
GtR
data.
And
if
anyone
ass
why,
say
you’re
hoping
to
put
an
entry
into
the
Visualising
Research
compe33on.
It’s
just
like
pugng
a
research
bid
in,
in
that
there’s
no
guarantee
of
a
payoff
(though
there
is
a
chance),
but
it’s
different
in
that:
a)
you’ll
have
learned
something;
b)
you’ll
have
already
have
finished
the
project
and
produced
some
hopefully
useful
output
to
meet
a
prespecified
need.
98
99. Here’s
an
example
of
a
typical
call
for
funding.
For
four
to
six
projects.
Up
to
£1.87
million
pounds.
I
hacked
around
the
food
data
space
for
a
couple
of
couple
of
hour
sessions
when
the
horsemeat
scandal
hit,
and
posted
a
couple
of
quick
blog
posts
(hHp://
www.open.edu/openlearn/science-‐maths-‐technology/compu3ng-‐and-‐ict/meat-‐here-‐
hun3ng-‐data-‐about-‐the-‐food-‐supply-‐chain
and
hHp://schoolofdata.org/2013/02/20/
made-‐to-‐measure-‐reshaping-‐horsemeat-‐importexport-‐data-‐to-‐fit-‐a-‐sankey-‐
diagram/
).
These
led
to
a
slot
at
an
Open
Data
Ins3tute
session
on
food
data:
hHp://
www.slideshare.net/psychemedia/odi-‐food
(full
annotated
slides
s3ll
to
follow...)
£1.87
million
pounds.
Four
to
six
projects.
99
100. Some3mes
you
have
a
scab
and
you
just
keep
on
picking
at
it..
Fun,
eh?
Dickens’
books
were
originally
produced
as
serials
in
popular
magazines
of
the
3me.
Many
of
today’s
longer
form
TV
series
are
wriHen
out
as
the
earlier
episodes
are
broadcast.
So
might
it
be
worth
thinking
about
guerrilla
research
as
a
form
of
serialised
research,
at
least
in
its
produc3on,
compared
to,
say,
a
Hollywood
blockbuster
film?
£1.87
million
pounds.
Sheesh…
100
101. Every
so
oeen
I
take
a
phrase
and
turn
it
somewhere
different.
Media
pluraility
refers
to
the
desire
to
have
the
ownership
of
the
apparatus
of
the
news
media
spread
across
several
(“a
plurality
of”)
different
owners.
News
is
content,
designed
to
inform
us
about
the
state
of
the
world.
News
is
oeen
published
from
a
par3cular
perspec3ve,
or
with
a
par3cular
slant.
The
editor’s
hand
is
always
there.
School
curricula
inform
us
too.
An
event
happened.
The
Times
reports
it
one
way,
the
Sun
another,
the
Guardian
yet
another.
The
na3onal
curriculum
is
published,
EdExcel
treat
it,
teach
it,
assess
it,
one
way,
AQA
another.
I
started
to
poke
around
looking
for
sta3s3cs
about
school
exam
sta3s3cs…
101
104. I
got
some
data,
and
had
a
play…
£1.87
million
pounds.
4
to
6
projects.
Sheesh…
104
105. Here’s
something
else
I’m
involved
with.
One
day
per
week
I
work
for
the
Open
Knowledge
Founda3on
on
a
project/ini3a3ve
called
the
School
of
Data.
The
School
of
Data
is
all
about
hands
on,
learning
by
doing
engagement
with
data.
The
audience
is
journalists
and
NGOs.
The
School
of
Data
do
a
thing,
a
really
neat
thing,
called
Data
Expedi3ons.
Get
a
topic,
a
group
of
people
with
an
interest
in
the
topic,
and
then
go
data
hun3ng;
frame
some
ques3ons
round
the
data
and
start
digging
in
to
it.
Look
for
stories
in
the
data,
then
find
a
way
of
telling
them.
In
a
day.
Or
less.
Or
over
a
week,
but
make
it
episodic.
Serialise
the
steps.
It
works,
too…
105
106. Means,
Opportunity
–
and
Mo3ve.
Why
bother?
Not
for
promo3on
[hHp://blog.ouseful.info/2010/08/26/in-‐for-‐a-‐penny-‐
in-‐for-‐a-‐pound-‐my-‐promo3on-‐case-‐for-‐support/
x
several
aHempts
so
far;
one
reason
I
dropped
to
4
days
per
week.].
So
why?
106
107. Because
it’s
fun…
and
maybe
because
it
could
be
useful.
Or
at
least,
interes3ng…
And
you
might
learn
something.
Or
beHer.
Like
how
to
do
something.
You
might
even
invent
how
to
do
that
something.
Or
innovate
a
solu3on
to
something
out
of
bits
and
pieces
that
already
exist.
107
111. There’s
a
lot
of
sensemaking
that
goes
on
in
the
world,
to
different
3mescales
and
budgets.
To
different
agendas.
For
different
purposes.
Channels
exist
between
these
different
communi3es,
conduits
that
pass
par3cular
sorts
of
informa3on,
or
impression,
packaged
in
par3cular
ways,
between
them.
Disrupt
the
f****rs.
111
112. What
sort
of
context
interests
you?
What
sort
of
context
is
important
to
you?
What
context
isn’t
working?
Can
you
cross-‐context?
Can
you
appropriate
a
context
and
use
it
to
jus3fy,
finesse
style,
your
guerrilla
research?
112
113. I
like
full
fact.
A
lot.
The
Conversa3on,
not
so
much.
But
the
Conversa3on
is
a
channel
that
some
universi3es
appear
to
support,
so
maybe
it’s
a
channel
you
can
use
to
provide
a
reverse
jus3fica3on
for
a
par3cular
piece
of
guerrilla
research,
par3cular
if
it
hooks
in
to
the
news
agenda.
113
114. “Why
are
you
doing
that?”
Response:
because
it’s
important.
114
115. Folk
keep
leaving
academia.
This
chap
lee
because
it
p****d
him
off
and
got
in
the
way.
So
he’s
building
a
solu3on
to
what
he
sees
as
part
of
the
problem.
Going
from
the
inside
to
the
outside
to
build
something
works
at
the
edge.
115
116. This
looks
interes3ng.
And
a
possible
jus3fica3on.
Do
something
low
risk
in
a
new
environment
to
experiment
with
a
new
workflow.
You
don’t
want
to
jeopardise
a
real
research
project
with
a
flaky
new
workflow,
aeer
all,
do
you,
really,
come
on?!
So
what
can
you
try
it
out
with…?
(The
tes3ng
of
the
environment
provides
the
jus3fying
context
for
what
you
do
inside
it…)
116
117. Exploring
workflows
that
embed
research
in
context,
exploring
tools
that
help
make
research
more
readable,
more
reproducible,
more
transparent,
seems
to
me
to
be
important
from
an
ed
tech
perspec3ve.
Notebook
style
working
works
for
me,
in
this
sense
[
hHp://blog.ouseful.info/2014/02/13/doodling-‐with-‐ipython-‐notebooks-‐for-‐
educa3on/
].
Have
you
tried
it
yet?
[
hHp://blog.ouseful.info/2014/02/26/3me-‐to-‐
drop-‐calculators-‐in-‐favour-‐of-‐notebook-‐programming/
]
Or
virtual
machines?
[
hHp://blog.ouseful.info/2013/12/02/packaging-‐soeware-‐for-‐
distance-‐learners-‐vms-‐101/
]
Use
either
as
context,
maybe,
for
some
guerrilla
research
of
your
own.
If
anyone
asks,
you’re
evalua3ng
the
notebook
way
of
working.
But
as
you
and
I
know,
that’s
also
to
provide
cover…
(Which
reminds
me:
have
I
men3oned
sabotage
yet…
or
corporate
foolery?
hHp://
blog.ouseful.info/2008/12/09/corporate-‐foolery-‐and-‐the-‐abilene-‐paradox/
)
By
the
by,
the
screen
shot
demonstrates
another
excuse
for
ac3vity.
Replica3ng
(and
in
this
case,
not)
a
piece
of
outstanding
work….
117
118. Another
top
tool
that
got
me
wondering
about
workflow.
Rstudio.
You
can
host
it:
hHp://blog.ouseful.info/2012/08/23/open-‐research-‐data-‐processes-‐
kmi-‐crunch-‐hosted-‐rstudio-‐analy3cs-‐studio/
And
it’s
a
gateway
drug
to
rapidly
prototypable
R
applica3ons:
hHp://
blog.ouseful.info/2013/06/21/disposable-‐visual-‐data-‐explorers-‐with-‐shiny-‐guardian-‐
university-‐tables-‐2014/
(you’ve
seen
this
before…)
118
119. Maybe
lack
of
knowledge
mo3vates
you…
Or
helping
come
up
with
ways
of
working
that
protect
us
from
ourselves…
119
120. But
that
won’t
stop
anyone…
As
a
guerrilla
researcher,
you
have
the
opportunity
to
do
just
as
much
and
just
as
valid
“research”
as
other
people,
because
they’re
making
it
up
too…
Only,
they’re
not
doing
it
for
a
reason…
They’ll
be
doing
it
because
it’s
their
job.
120
121. There’s
work
to
be
done.
This
is
one
of
the
ques3ons
that
drives
me.
How
to
make
use
of
all
the
stuff
that’s
out
there
already?
How
to
put
it
together
so
it
works
together?
How
to
use
one
bit
that
exists
to
help
make
sense
of
another
bit
that
exists.
Outside
one
of
the
mee3ng
rooms
at
the
OU,
there
is,
or
at
least
was,
a
framed
jigsaw
on
the
wall.
The
picture
seemed
to
make
sense.
Or
maybe
it
didn’t.
Because
that
jigsaw
was
made
from
pieces
from
different
jigsaws.
I
liked
that.
It
made
sense.
Par3cularly
in
that
environment.
Different
people,
coming
together
with
independent
ideas,
leaving
with
the
same
picture.
What
drives
you?
121
122. Last
tool.
Last
toy.
This
is
Gephi
[
hHp://gephi.org
]
–
a
cross-‐playorm
desktop
tool
that’s
great
for
genera3ng
effec3ve
network
visualisa3ons.
I
have
some
tutorials
and
sample
datasets
if
anyone
wants
to
give
it
a
whirl…[
hHp://blog.ouseful.info/2012/11/09/drug-‐deal-‐
network-‐analysis-‐with-‐gephi-‐tutorial/
Or
do
some
guerrilla
research
around
your
Facebook
network
by
googling
this:
site:blog.ouseful.info
in7tle:"facebook
network”
Alterna3vely,
see
if
what
they
like
reveals
anything
about
you…
hHp://
blog.ouseful.info/2012/01/04/social-‐interest-‐posi3oning-‐visualising-‐facebook-‐friends-‐
likes/
]
122
123. I
like
networks.
A
lot.
I
like
them
as
maps.
Maps
help
you
make
sense
of
a
space,
help
you
navigate
a
space.
They
give
you
a
view
over
the
whole,
over
the
parts,
over
how
the
parts
relate
to
each,
how
the
parts
relate
to
the
whole,
how
the
whole
relates
to
the
parts.
At
least,
in
part.
This
map
shows
where
I
posi3on
myself,
or
at
least,
where
I
am
socially
posi3oned,
on
TwiHer.
It’s
based
on
how
my
followers
follow
each
other
on
TwiHer,
grabbed
some
3me
ago.
Gephi
drew
it,
with
my
help.
Or
maybe
I
drew
it,
with
Gephi’s
help.
You
can
draw
things
like
this
too…
in
the
simplest
case
all
you
need
is
two
columns
of
data,
from
and
to.
A
two
column
CSV
file.
Each
line
says:
draw
a
line
from
from
to
to.
And
Gephi
will.
Then
you
can
place
the
points,
with
Gephi’s
help.
Or
Gephi
will
place
the
points,
with
your
help.
I’m
never
really
sure
which.
I
can
see
territories
in
my
map,
because
the
names
are
meaningful
to
me.
Each
name
has
an
associa3on,
in
interest
space,
what
to
me
are
the
interests
of
each
TwiHer
user
displayed.
Together,
their
interests
coalesce.
The
colours
help
reinforce
that.
The
map
has
regions.
It
makes
some
sort
of
sense,
the
sense
made
by
the
sense
of
each
point
on
the
map,
and
the
interests
they
express
whenever
they
make
a
connec3on
to
another
person
on
TwiHer.
The
map
of
my
interests
is
beyond
my
control
–
it’s
a
map
of
my
interests
as
projected
by
the
interests
of
people
who
follow
me
on
TwiHer.
We
can
draw
other
maps
too.
I
par3cularly
like
emergent
social
posi7oning
maps,
projec3ons
from
the
followers
of
an
individual,
or
users
of
a
hashtag,
onto
the
people
123
124. Once
you
start
looking,
you
find
opportuni3es
to
grab
graph
data,
edge
data,
connec3on
data,
from
all
sorts
of
places.
Here’s
an
example
built
out
of
the
Shell
corporate
sprawl.
The
data
comes
from
opencorporates.com,
in
the
form
of
directors
associated
with
companies.
I
draw
lines
between
companies
and
directors,
then
remove
the
directors
and
add
edges
to
connect
companies
that
shared
two
or
more
directors.
The
labels
are
sized
rela3ve
to
the
PageRank
score
of
each
node,
which
a
measure
of
how
well
connected
the
node
is
in
the
graph
(the
“importance”
of
each
node
is
dependent
on
the
“importance”
of
the
nodes
connected
to
it….)
The
lines
also
provide
a
background
that
highlights
the
connec3vity
-‐
and
structure
–
of
the
corporate
elements.
There’s
a
recipe
here
-‐
hHp://www.slideshare.net/psychemedia/school-‐of-‐data-‐
mapping-‐company-‐networks
–
for
working
with
OpenCorporates
data
that
also
makes
use
of
OpenRefine
and
Gephi.
I
had
to
work
out
how
to
do
it
myself,
but
you
can
follow
along
if
you
want
to…
Or
you
can
start
off
by
following
my
way,
then
make
up
your
own.
Or
just
go
for
it.
JFDI.
124