Are You Just
‘Organising It’?

Part 1

Jon Reades - j.reades@ucl.ac.uk

22nd October 2026

Last Week, You Said…

It is because we are collecting exponentially more massive amounts of data and so the tools and methods must adapt to the scale

Data science builds upon decades (maybe longer – knitting or cave paintings, I would argue, are a form of data science) of progress […]

Practicing working with different datasets that are not perfect.

There were 17 Week 2 submissions. Not much feedback.

These are anonymous, and this is what happens to them. So write one at the end today.

About 13 of 17 said “not new, but more data and more computing”, which is close to Donoho’s “coping skills”. Only a few said “new” (the first quote is the clearest). Almost nobody cited a reading by name!

Why these three?

• Scale: see the GDELT example.
• Knitting and cave paintings: this is Mattern’s point.
• “Not perfect”: who decided what ‘perfect’ looks like?

On learning, “practice” was by far the most common answer. One said “Not relying on AI to write code for you”.

Maybe git commits are a record of your own practice?

This Week’s Question

When you decide how to store data, are you just organising it, or already deciding what it means?

 

We’ll answer it twice: with Python, with cities.

Asking a group worked for diversifying feedback. So don’t ask for volunteers: walk to a row and ask “What did you two decide?” The answer is then “we”, not “I”.

In the DOLs video I say there’s “no right way” to store data, then, thirty seconds later, that there is a right answer. Today is about which of those is true.

Predict the Output

Pairs. No laptops. Vote with your fingers.

Pairs means whoever is sitting next to you.

1. Deliberately Similar

cities = [
  'San Francisco',
  'London',
  'Paris',
  'Beijing']

# Prints London
print(cities[2])

1. San Francisco
2. London
3. Paris
4. IndexError

Answer: 3. Indexes start at 0. And the comment is wrong: it’s copied from my own slide in Dictionaries.

From Dictionaries (“Deliberately Similar”). The point: a comment is context, and context can be wrong. Remember that for the readings.

Story (≤ 60 seconds): why a dictionary key has to be immutable using example of zebras relabelled as finches.

2. Find the 5

my_list = [
  [1, 2, 3],
  [4, 5, 6],
  [7, 8, 9]
]

Which prints 5?

1. my_list[1][1]
2. my_list[2][2]
3. my_list[1,1]
4. my_list[4]

Answer: 1. Row 1, column 1, counting from zero. (2 gives 9; 3 is a TypeError; 4 is an IndexError.)

From LOLs. Story if time: a list-of-lists is like the reference list in a journal article. It points to other things rather than containing them.

But what does my_list actually look like if you take a step back and squint? Answer: a matrix or a data set!

3. How Is That Easier?

ds2 = {
  'lat': [51.51,40.71,35.69],
  'lon': [0.13,74.01,139.68],
  'tz':  [+0,-5,+9],
  'name':['London','New York','Tokyo']
}

print(ds2['tz'][ ds2['name'].index('Tokyo') ])

1. 0
2. -5
3. 9
4. KeyError

Answer: 3. But is there another issue?

From DOLs to Data.

Story: it took trying to teach pandas for me to really get my head around why this structure is so good.

Don’t explain the +9 yet: ask “How many of you noticed?” (expect almost no hands) and say that we’ll come back to it after the break (3.7).

Just Organising It?

Vote

According to the reading you did, when we decide how to store data, we are…

 

1. Just organising it: an efficiency choice.
2. Mostly organising it, but the structure decides which questions can be asked.
3. Already deciding what it means: making choices about what counts.

Who read: D’Ignazio & Klein? Mattern? Both? Neither? (Hands!)

Then vote by reading: D’Ignazio readers first, then Mattern.

Then 1 minute of silent writing (“my vote, and one passage”), then 3 minutes with neighbours: compare votes and find one passage that proves yours. Anyone who didn’t read is briefed by a neighbour. Writing first gives quieter students something to say.

Plenary (15 min): start with where each reading would send you to look for the missing context (see “Same Diagnosis, Different Cure?”). Walk the room and ask pairs, not individuals; ask which reading they used only after they’ve answered (“Was that a D’Ignazio pair?”).

Counterpoints from margin notes (use one):

• D’Ignazio p.6 (the procurement-data questions): DOGE used crude keyword filters to find ‘woke’ research and cut it, without finding out what the research was about or how procurement and research funding work.
• D’Ignazio p.2 (“Consider Context”): this connects to an AI’s context window. How was it trained? What’s in the frame?
• Mattern p.1 (Sidewalk Labs): it failed in Toronto. The pitch was to ‘start up’ a new part of the city without ‘legacy’ infrastructure.
  • TVO Report
  • Governing.com article
  • Chapter from Ineffective Policies
  • Regional Studies article

D’Ignazio & Klein (2020)

Although GDELT de-duplicated some of those stories to a single event, it still logged, erroneously, that hundreds of kidnapping events had happened that day. ~ D’Ignazio and Klein (2020, 2)

[…] data enter into research projects already fully cooked—the result of a complex set of social, political, and historical circumstances. ~ D’Ignazio and Klein (2020, 13)

The first quote is about FiveThirtyEight’s use of GDELT after the Chibok kidnapping: media reports were counted as events.

“Fully cooked” is D’Ignazio and Klein drawing on Gitelman and Jackson, whose line is “‘Raw data’ is an oxymoron”.

My margin notes:

• p.2 (FiveThirtyEight): “The problem of finding ‘interesting data’ and then running with it. Especially if on a deadline.”
• p.3 (GDELT’s hype): “This data isn’t coming from nowhere, it’s coming from people taking decisions about what to log, and what not to log, how to log it, and how to organise it.” That is this week’s question.
• p.13: “All ‘error is political’” (see the week 1 closer).

Mattern (2017)

Our current paradigm, the city as computer, appeals because it frames the messiness of urban life as programmable and subject to rational order. ~ Mattern (2017)

Places Journal has no page numbers.

My margin notes:

• p.2: “If you translate the city into a computer then you proceed on the basis of that understanding and become less able to recognise where your assumptions are incorrect.”
• p.3 (the second quote): “Key point of the essay. Metaphors shape our understanding!”

It matters because the metaphors give rise to technical models, which inform design processes, which in turn shape knowledges and politics, not to mention material cities. ~ Mattern (2017)

My margin notes:

• p.1 (“What should a city optimize for?”): “Optimisation assumes some kind of objective function for what ‘matters’.” That is the language of machine learning leaking into governance.

Link back to the videos: at the end of LOLs I say that “Excel has been presenting everything to us the wrong way round”. Metaphors shape code too.

Same Diagnosis, Different Cure?

Compare the perspectives of:

But deep context and computation are not incompatible. ~ D’Ignazio and Klein (2020, 16)

And:

City-making is always, simultaneously, an enactment of city-knowing — which cannot be reduced to computation. ~ Mattern (2017)

On the week’s question they’re on the same side: both would vote 3. The difference is in the ‘cure’, not the diagnosis.

The plenary’s opening question: “Where would each of them send you to find the missing context?”

• D’Ignazio and Klein: inside data science. Go back to whoever collected the data, the domain experts, and the data’s history (the data biography). Computation done better.
• Mattern: outside the database. Librarians, archives, residents, maintenance, and knowledge from the senses and long familiarity with a place. She questions the ‘city as computer’ frame itself, though she doesn’t reject data (“and, yes, by systematically filtering data”).

Don’t invent a fight: this is a difference of emphasis. Indeed, I think Catherine and Lauren would actually agree with Shannon on this, but in this particular chapter they put the emphasis in different places. Indeed, D’Ignazio and Klein cite Mattern on librarians being left out of smart-city conversations (p.26).

My margin note: “Why these are paired.”

Then straight into 3.6 (Group Working); break comes after.

Thank You

References

D’Ignazio, Catherine, and Lauren F. Klein. 2020. “Data Feminism.” In. MIT Press. https://data-feminism.mitpress.mit.edu/.

Mattern, Shannon. 2017. “A City Is Not a Computer.” Places Journal. https://doi.org/10.22269/170207.