Is Data Science New?

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

13th October 2026

This Week’s Question

Is data science new? And, if so, how you should learn it?

 

• The readings answer the first half.
• On Writing & Coding answers the second half.

Run sheet (about 85 minutes before handing over to 2.7):

• The deal: 5 min
• Predict the output: 20 min
• Is data science new?: 35 min (including neighbour discussion)
• Break: 10 min
• What does a data scientist do?: 15 min
• Bridge to On Writing & Coding

Voting is with fingers (1–4) unless Mentimeter is open in a second browser tab. Pairs means whoever is sitting next to you.

The Deal

[Tukey] saw apprenticeship with real data analysts and hence real data as the solution. ~ Donoho (2017, 754)

From now on, I assume that you’ve watched the videos and done the reading. We will use them, not repeat them.

My margin note on this page says: “This is what I’m trying to do with FSDS! Real problems. Real experience.”

This is the justification for the format: you can’t apprentice by watching someone else lecture. Class is where we use what you’ve prepared.

Predict the Output

Pairs. No laptops. Vote with your fingers.

1. Strings

w, x, y, z = '4a','4a','365','42'
print(x > y)
print(x > z)

 

1. True False
2. True True
3. False False
4. An error

Answer: 2. Strings are compared character by character, and '4' > '3'; then 'a' > '2'.

From Basics 1. Story (≤ 60 seconds): Point: if you’re not careful about what type of data you’re working with, you get odd, hard-to-diagnose bugs. Tie-in: Grace Hopper’s moth.

2. Slicing

geographers = ["Massey", "Harvey", "Rose"]
i = 2
print( geographers[ (i-3)**2-4:-1 ] )

 

1. ['Massey', 'Harvey']
2. ['Harvey', 'Rose']
3. ['Rose']
4. An error

Answer: 1. (2-3)**2-4 is -3, so this is geographers[-3:-1].

From Lists (“Test Yourself”). Ask a pair who got it right to talk the room through the arithmetic. Work it inside out, the same way Python does.

3. Appending

female = ['Rose','Valentine','Massey','Jefferson']
male   = ['Von Humboldt','Harvey','Hägerstrand']
a = female + male
b = []
b.append(female)
b.append(male)
print(len(a), len(b))

Answer: 7 2. + joins two lists; append adds one thing, which here happens to be a list.

Why does append look inconsistent? It isn’t inconsistent, it does one thing every time.

Story: Basics 2, deleting code I didn’t understand.

4. Tricksy

Same output. How are they different?

geographers = ['Rose','Massey','Jefferson']
geographers.reverse()
for g in geographers:
  print(g)

Jefferson
Massey
Rose

geographers = ['Rose','Massey','Jefferson']
g = len(geographers)-1
while g >= 0:
  print( geographers[g] )
  g -= 1

Jefferson
Massey
Rose

reverse() changes the list in place: after the left-hand version, geographers is still backwards.

Link back to sort() in Lists.

Two pieces of code that give the same output are not the same code. That matters later, when the next line depends on the order of the list.

Is Data Science New?

Vote

For reading you did, what was its verdict on data science (or quantitative geography)?

 

1. New: something genuinely different.
2. Rebrand: old work under a new name.
3. Bandwagon: driven by fashion, money or administrators.

Who read: Burton? Donoho? Franklin? None?

Then vote by reading: Burton readers, hand up. Then Donoho, then Franklin. Note where the readings disagree.

Then 5 minutes with neighbours: compare verdicts and find one passage that proves yours. Anyone who didn’t read gets briefed by a neighbour.

Plenary: start where the readings disagree, and cold-call by reading (“Franklin readers, who said rebrand? Why?”).

Burton (1963)

It is no flight of fancy to foresee the day when geographers, if they are to remain abreast of developments, must re-learn their craft anew every decade. ~ Burton (1963, 153)

If the revolution had been inspired by belief in quantification for its own sake, or by fad and fashion, then it would have rapidly run its course and quickly died. ~ Burton (1963, 156)

Burton’s verdict: the revolution is over because its ideas became conventional, and it lasted because it had a purpose (theory), not because it was fashionable.

“Re-learn their craft anew every decade”: that was written in 1963. That’s you now, with LLMs.

Donoho (2017)

Statisticians see administrators touting, as new, activities that statisticians have already been pursuing daily, for their entire careers […] ~ Donoho (2017, 746)

The new skills attracting so much media attention are not skills for better solving the real problem of inference from data; they are coping skills for dealing with organizational artifacts of large-scale cluster computing. ~ Donoho (2017, 747)

My margin notes: “Note the academic ‘bandwagon’ that everyone wants to jump on, esp. administrators.” And on p.4: “New ‘skills’ are a function of the hardware available.”

Counterpoint for the room: is this any different from the way CS and data science have moved into geography and planning?

Franklin (2024)

Gone, too, is the illusion of a monolithic, unified quantitative human geography, replaced by urban analysts, geographic data scientists, geospatial data scientists, and GeoAI practitioners. ~ Franklin (2024, 237)

By identifying with tools and analytics, quantitative methods effectively shunts itself even further away from human geography, pigeonholing quantitative human geographers as methods people. ~ Franklin (2024, 241)

My margin note on p.6: “Key problem here!” Franklin’s worry isn’t whether the work is new, it’s what happens to identity when people name themselves after their tools.

So… Nothing New Under the Sun?

Four major influences act on data analysis today:

1. The formal theories of statistics
2. Accelerating developments in computers and display devices
3. The challenge, in many fields, of more and ever larger bodies of data
4. The emphasis on quantification in an ever wider variety of disciplines

What year?

Tukey, 1962 (quoted in Donoho (2017, 749)).

Ask the room to guess the year before revealing it. My margin note: “Pretty much sounds like where we are today as well!”

What Does a (Spatial) Data Scientist Do?

Put These in Order

• Code
• Tidy
• Analyse
• Understand
• Communicate
• Report

Pairs, 2 minutes: put these in order. Take two or three orders from the room.

This order is how to get a mark of between 45 and 63 in your dissertation.

A Better Order

1. Understand
2. Tidy
3. Code
4. Report
5. Analyse
6. Communicate

Not this…	…but this
How should we do this?	Why are we doing this?
This is the number.	These are the takeaways.
These are the methods I can use.	This is the method that matches the need.
What do I need to say?	Who needs to know?

 

Quantitative analysis of variables cannot be justified for its own sake. ~ Burton (1963, 158)

It’s also a loop: each step feeds back into the others. Reports are for you (diagnostics, data quality) before they’re for anyone else.

Burton’s line is “Understand” put another way: 60 years ago, the same argument.

Compare: Donoho’s ‘Greater Data Science’

1. Data gathering, preparation, and exploration
2. Data representation and transformation
3. Computing with data
4. Data modeling
5. Data visualization and presentation
6. Science about data science

Neighbours, 4 minutes: what does each list leave out? Take three or four answers.

(Donoho’s own section headings swap 4 and 5; this is the order of his summary list.)

Mine has Understand and Communicate; his has science about data science and, implicitly, reproducibility. Neither has “use the latest tool”.

Why the Practicals Look Like This

Students who only analyze precooked data are not being given the chance to learn these essential skills. ~ Donoho (2017, 757)

An article about computational science in a scientific publication is not the scholarship itself, it is merely advertising of the scholarship. ~ Donoho (2017, 760), summarising Jon Claerbout

This is why the practicals use real, messy data, and why the assessments are submitted as reproducible notebooks: the code is the scholarship.

The Story is Stable

Job titles keep changing. Here’s one version of that from Analyst Uttam (2025):

Old Data Science Role	New Job Title
Data cleaning & ETL	Analytics Engineer (SQL + dbt + Airflow)
Business dashboards	BI Developer (Power BI, Tableau, Looker)
Product experimentation	Data Analyst (SQL + experimentation)
ML Modelling	ML Engineer (MLOps, pipelines, deployment)
Causal inference/forecasting	Research Scientist (Stats + PhD-level modeling)

Back to Franklin: if you identify with the tool, you’re pigeonholed when the tool changes. These titles apply in the private sector; academics tend to be one-stop shops. The point isn’t that all these roles exist, it’s that the way the work is divided keeps changing.

So What Is Today’s ‘Coping Skill’?

The new skills attracting so much media attention […] are coping skills. ~ Donoho (2017, 747)

If Donoho’s ‘coping skills’ were about dealing with “the severe new constraints on algorithms posed by the multiprocess/networked wolrd”, what’s ours?

Take one or two answers (someone will say LLMs), then switch to On Writing & Coding.

Thank You

References

Analyst Uttam. 2025. “‘Data Science Is a Dead Career’ — the Truth Behind the Trend No One Wants to Say Out Loud.” https://medium.com/ai-analytics-diaries/https-analystuttam-substack-com-p-data-science-is-a-dead-career-87ee2d8bd338.

Burton, I. 1963. “The Quantitative Revolution and Theoretical Geography.” The Canadian Geographer/Le Géographe Canadien 7 (4):151–62. https://doi.org/10.1111/j.1541-0064.1963.tb00796.x.

Donoho, D. 2017. “50 Years of Data Science.” Journal of Computational and Graphical Statistics 26 (4):745–66. https://doi.org/10.1007/978-3-642-23430-9_71.

Franklin, Rachel. 2024. “Quantitative methods III: Strength in numbers?” Progress in Human Geography 48 (2). SAGE Publications Sage UK: London, England:236–44. https://doi.org/10.1177/03091325231210512.