5. Either of these sheep could be a ram, which,
according to some definitions, is different from
a sheep. But you can't tell by looking.
Similarly, either could be a female sheep. And a
female sheep might be pregnant.
Assuming sheep only have one baby sheep at
a time (they don't, but anyhow), how many
sheep is a fetus?
Is it one sheep, half a sheep, zero sheep?
6. So depending on how we define 'sheep', the
number of sheep in the photo might be...
0
1
1
1.5
2
2
2
2
2
2.5
2.5 + 0.5
3
4
7. So depending on how we define 'sheep', the
number of sheep in the photo might be...
0 Both rams
1 One ram, one female sheep
1 One ram, one pregnant female sheep - the fetus = 0 sheep
1.5 One ram, one pregnant female sheep - the fetus = 0.5 sheep
2 Two animals of species 'sheep'
2 Two female sheep
2 One ram, one pregnant female sheep - the fetus = 1 sheep
2 Two female sheep, one pregnant - the fetus = 0 sheep
2 Two female sheep, both pregnant - the fetuses = 0 sheep
2.5 Two female sheep, one pregnant - the fetus = 0.5 sheep
2.5 + 0.5 Two female sheep, both pregnant) (disequals '3 sheep')
3 Two female sheep, one pregnant - the fetus = 1 sheep
4 Two female sheep, both pregnant - the fetus' = 1 sheep each
8. ...so as we can see, when tinkering with data, a
LOT depends on how we define the terms.
9. ...so as we can see, when tinkering with data, a
LOT depends on how we define the terms.
A lot also depends on the depth of the data we
have to work with: we know that these two
animals will be either
- female or male
- pregnant or not pregnant
...but the (visual) data provided by the photo
can't tell us which is which.
10. Data can be very misleading, too: if someone
tells you authoritatively 'there are two sheep in
the field', they're probably right.
But which two sheep do they mean?
2 Two animals of species 'sheep'
2 Two female sheep
2 One ram, one pregnant female sheep - the fetus = 1 sheep
2 Two female sheep, one pregnant - the fetus = 0 sheep
2 Two female sheep, both pregnant - the fetuses = 0 sheep
...and could they inadvertently be counting
different types of 'two sheep' the same way?
12. What's the point of sheep?
Sheep are cute and fluffy and fun to chase.
But when we collect data on sheep, we're
usually doing so for a specific purpose.
Like, for example, getting a better wool yield.
13.
14. There are lots of ways of increasing wool yield:
- giving the sheep better grazing pastures
- controlling their parasites
- vaccinating them
15. ...but not weighing them.
You can't make a sheep produce more wool by
weighing it.
16. Weighing sheep is necessary if you want data
about sheep weight.
But data by itself doesn't actually do anything.
17. Weighing sheep is necessary if you want data
about sheep weight.
But data by itself doesn't actually do anything.
When the data is coaxed, crunched, and ultimately
transformed into information, it becomes useful...
18. Weighing is necessary if you want data about
sheep weight.
But data by itself doesn't actually do anything.
When the data is coaxed, crunched, and ultimately
transformed into information, it becomes useful...
But 'being useful' is not the same as 'being used'.
19. Weighing is necessary if you want data about
sheep weight.
But data by itself doesn't actually do anything.
When the data is coaxed, crunched, and ultimately
transformed into information, it becomes useful...
But 'being useful' is not the same as 'being used'.
Data is not an end in itself.
Neither is information.
20. The point where data finally becomes useful is
when we take the insights from the information
we've learned and apply it back to the thing we
want to improve.
21. The point where data finally becomes useful is
when we take the insights from the information
we've learned and apply it back to the thing we
want to improve.
Data
Information
Change in practice
Better outcomes
23. (Now replace 'sheep' with 'British students'
and 'wool yield' with 'better A level results
in Maths'.)
(You get the idea.)
24. This is all well and good, but clearly, getting
good, useful information out of data so that
people can use it to make things better is not
an easy task.
It can be time consuming, highly skilled, and
expensive.
So who is going to do the work?
25. If you give us data to play with,
Developers will! Developers are lovely people who like to
spend their time making things that they think will help
other people.
Some of us do it for fun. Some of us do it for big
companies. And some of us build entire SMEs out of
making data useful!
we will do all the
work!
28. "Transparency Drives
Prosperity"
...now who said that?
Open Data White Paper: Unleashing the
Potential (28 June 2012)
http://www.cabinetoffice.gov.uk/resource-
library/open-data-white-paper-unleashing-
potential
30. Originally presented by Zoë Rose
at Young Rewired State open data hack day, 7 July 2012 http://hacks.rewiredstate.org/events/npd2012
(Basically I didn't have anything built that actually worked, so I showed the audience - including
Michael Gove - some pictures of sheep instead.)
Notas do Editor
How many sheep in this picture?
Easy - there are two, right?
Well, maybe...
Either of these sheep could be a ram, which, according to some definitions, is different from a sheep. But you can't tell by looking.
Similiarly, either could be a ewe. And a ewe might be pregnant. Assuming sheep only have one baby sheep at a time (they don't, but anyhow), how many sheep is a fetus? Is it one sheep, half a sheep, no sheep? And what if the fetus is a ram?
So depending on how we define 'sheep', the number of sheep in the photo might be...
0
1
1.5
2
2
2
2.5
2.5 + 0.5
3
4
So depending on how we define 'sheep', the number of sheep in the photo might be...
0 (both rams)
1 (one ram, one female sheep)
1 (one ram, one pregnant female sheep - but we don't count the fetus)
1.5 (one ram, one pregnant female sheep - the fetus = 0.5 sheep)
2 (two animals of species 'sheep')
2 (two female sheep)
2 (One ram, one pregnant female sheep - the fetus = 1 sheep)
2 (Two female sheep, one of which is pregnant - the fetus = 0 sheep)
2 (Two female sheep, both of which are pregnant - the fetuses = 0 sheep)
2.5 (Two female sheep, one of which is pregnant - the fetus = 0.5 sheep)
2.5 + 0.5 (Two female sheep, both of which are pregnant) (note: not the same as '3 sheep')
3 (Two female sheep, one of which is pregnant - the fetus = 1 sheep)
4 (Two female sheep, both of which are pregnant - the fetus' = 1 sheep each)
...so as we can see, when tinkering with data, a LOT depends on how we define the terms.
A lot also depends on the depth of the data we have to work with - we know that these two animals will be either female or male, and either pregnant or not pregnant, but the data provided by the photo can't tell us which is which.
Data can be dangerous this way: if someone tells you 'there are two sheep in the photo', it's easy to conclude that it's a simple answer.
Data can be dangerous this way: if someone tells you authoritatively 'there are two sheep in the field', it's easy to conclude that they're right.
But which two sheep do they mean?
2 (two animals of species 'sheep')
2 (two female sheep)
2 (One ram, one pregnant female sheep - the fetus = 1 sheep)
2 (Two female sheep, one of which is pregnant - the fetus = 0 sheep)
2 (Two female sheep, both of which are pregnant - the fetuses = 0 sheep)
Sheep: what's the point?
Sure, sheep are fun and cute and make adorable noises. But when we collect data on sheep, we're usually doing so for a specific purpose - like, for example, getting a better yield of wool.
There are lots of ways of increasing wool yield - giving the sheep better grazing pastures, controlling their parasites, vaccinating them...
...but not weighing them. You can't make a sheep grow more wool by weighing it.
Weighing is necessary to get data. But data by itself doesn't actually do anything.
When the data is coaxed, crunched, and ultimately transformed into information, it becomes useful...
But even then, it still doesn't actually do anything! It's just a chart/report/graph/etc.
Collecting data is not an end in itself. Converting data to information is not an end in itself either.
The point where data finally becomes useful is when we take the insights from the information we've learned and apply it back to the sheep. That's how we can get an enhanced wool yield.
Data -> Information -> Change in practice -> Better outcomes
(Now replace 'sheep' with 'British students' and 'wool yield' with 'better A level results in Maths'.)
But making sense of data is a huge job - Who will do it?
We will!