The DIG Manifesto
Fabian’s and Albert’s interpretation of
© Fabian M. Suchanek and Albert Bifet
Context
• NewUni
“évoluer vers un MIT [...] à la française” [Convention de Coop. NewUni]
2
• NewUni
“évoluer vers un MIT [...] à la française” [Convention de Coop. NewUni]
• ParisTech
“une grande ambition de MIT à la française” [
Le Monde, 2013-06-06
]
Mais: “abandon (voulu par l’État) de toute ambition institutionnelle
pour ParisTech, [un] échec réel” (Y.P. 2016-09-23)
3
Context
• ParisTech
“une grande ambition de MIT à la française” [
Le Monde, 2013-06-06
]
Mais: “abandon (voulu par l’État) de toute ambition institutionnelle
pour ParisTech, [un] échec réel” (Y.P. 2016-09-23)
• NewUni
“évoluer vers un MIT [...] à la française” [Convention de Coop. NewUni]
• Institut Mines‐Télécom
“L’IMT, c’est un peu comme le MIT à la française”
[Patrick Duvaut to the Romanian ambassador, 2018-06-28]
But: the IMT has disappeared from our publications.
[P.D. on 2015-10-05]
4
Context
• Paris‐Saclay
“reach the same level of excellence as [the] MIT” [
The Conversation
]
But: the ambition decreases steadily
10
2013 2014 2015 2016 2017
Top 8
Top 10
Top 15
Top 20
Top 30
20
30
40
50
ｎ
Wait a minute: how does the MIT actually look?
Paris-Saclay’s
self-proclaimed
goal of being
a top ｎ university
5
Context
The DIG Manifesto
But we want to make a point:
Take the international universities as a role model
not just in their ranking, but also in their inner workings.
6
We have not studied at the MIT, but we have studied, worked at
or taught at the Max Planck Institute, La Sapienza, Waikato U,
Barcelona U, and Oxford U.
We want NewUni to succeed, and we are committed to
supporting the new institution.
(and unlike JLC, we will be around for the years to come)
(INTRO, INT, CONC)
“Il faut penser de l’excellence pas comme objectif,
mais comme résultat.” — Valérie Beaudouin
This Talk
7
This talk points out
• how an ideal university on the international level would look.
• what currently holds us back.
Overview:
•
Challenge 1: Institutional complexity
•
Challenge 2: Too many verticals
•
Challenge 3: Infrastructure
•
Challenge 4: Shared Calendar
•
Challenge 5: Research Organization
•
Challenge 6: Master’s Programs
(Fabian’s and Albert’s position, not DIG Manifesto)
The slides refer in darkred capital letters to the item names of the
DIG Manifesto or the GoodUni Manifesto, respectively.
Challenge 1: Institutional Complexity
Télécom
Mines
Eurecom
Paris-Saclay
School SOEIST
INFRES
IC2
DBWEB
DIG
LIX
INRIA
LINCS
CRG
CERNA
SES
I3
Joint Research
Program on Big
Data Management
AxSE BigData
SSA
STA
TSI
IDS
Pôle DCA
OAK2
CEDAR
TUM
Franco-
German
Academy
LTCI
LabEx
DigiCosme
Polytechnique
ParisTech
Dép. STIC
Axe Big Data
gray: obsolete
red: renamed
Mines-Télécom
Mines-Ponts
CNRS
8
Télécom
Evolution
Télécom
Mines
Eurecom
Paris-Saclay
School SOEIST
INFRES
IC2
DBWEB
DIG
LIX
INRIA
LINCS
CRG
CERNA
SES
I3
Joint Research
Program on Big
Data Management
AxSE BigData
SSA
STA
TSI
IDS
Pôle DCA
OAK2
CEDAR
TUM
Franco-
German
Academy
LTCI
LabEx
DigiCosme
Polytechnique
ParisTech
Dép. STIC
Axe Big Data
gray: obsolete
red: renamed
Mines-Télécom
Mines-Ponts
CNRS
NewUni
DS Faculty
This is not how the MIT looks!
9
Challenge 1: Institutional Complexity
Télécom
Evolution
institutional changes
(CNRS, NewUni, Saclay, TSP)
students
deaths & recruitments
DSI
terror attacks
Mails received from Y.P. via DiffZ since 2013:
Institutional complexity
• complicates our lives
• consumes an enormeous amount of time
10
Challenge 1: Institutional Complexity
Institutional complexity
• complicates our lives
• consumes an enormeous amount of time
• makes us look ridiculous
11
Challenge 1: Institutional Complexity
Wikipedia/MIT
Wikipedia/Telecom
Institutional complexity
• complicates our lives
• consumes an enormeous amount of time
• makes us look ridiculous
• costs us visibility
12
Challenge 1: Institutional Complexity
SCOPUS knows
279 name variants
of Télécom ParisTech
You think it’s all spelling variants? That’s not true, see next slide.
Every name variant costs us visibility!
Institutional complexity
• complicates our lives
• consumes an enormeous amount of time
• makes us look ridiculous
• costs us visibility
13
Challenge 1: Institutional Complexity
Publication signatures of a single Télécom ParisTech researcher over the past 5 years:
11 signatures in 5 years = more than 2 per year!
Our Proposal
14
Télécom
ENSTA
Polytechnique
ENSAE
TSP
IMT
Paris-Saclay
ParisTech
T. Evolution
1. Recognize institutional complexity as a problem that has to be solved.
Our Proposal
15
NewUni
IMT
Paris-Saclay
ParisTech
1. Recognize institutional complexity as a problem that has to be solved.
2. Merge the NewUni schools (or have this as a longterm goal,
or merge as much administration and layers as we can)
If we don’t want to merge, that’s fine! Then just stop comparing
the projet with “MIT”, “Caltech” or “EPFL”.
(MERGE)
“encore un autre
Paris-Saclay ou IMT”
T. Evolution
Our Proposal
16
NewUni
1. Recognize institutional complexity as a problem that has to be solved.
2. Merge the NewUni schools (or have this as a longterm goal)
3. Quit all the other institutions
(MERGE, QUIT)
“Ce qui est clair et limpide nous attire,
ce qui est embrouillé nous rebute.”
— David Hilbert
Our Proposal
17
NewUni
1. Recognize institutional complexity as a problem that has to be solved.
2. Merge the NewUni schools (or have this as a longterm goal)
3. Quit all the other institutions
4. Use a single signature for publications: NewUni
Advantages:
• simplification
• synergies
• reduction of costs
• gain in visibility
(MERGE, QUIT, VISI)
Challenge 2: Too many verticals
18
The Lab coordinates
my research, has
• budget
• label
• council
• director
“les resp de dept qui gèrent l’enseignement et la recherche” [Y.P. 2018-05-07]
The department coordinates my
teaching and my research, has:
• budget
• label
• council
• director
19
The Lab coordinates
my research, has
• budget
• label
• council
• director
The department coordinates my
teaching and my research, has:
• budget
• label
• council
• director
Problems:
• unclear responsibilities concerning research, potential for conflict
• overhead with budget, label, councils on both sides
• international universities do not have separate structures
for teaching and research (with separate budgets, labels, etc.)
Fun fact: I have never talked with either of them
about my teaching or my research (and that’s OK).
Challenge 2: Too many verticals
Our Proposal
20
Research strategy:
by the director of research
Teaching strategy:
by the director of teaching
and/or a committee
of the department directors
This is how international universities work:
no separate structures for teaching and research.
At the level of the school, we should have
• budget (shared budget)
• council (with the heads of the groups + directors)
(ONESTRUC)
Our Proposal
21
School/Faculty/Department
• budget (shared budget)
• council (with the heads of the groups + directors)
NewUni
Group
Individual
(3LAYER)
22
Paris-Saclay admitted the students without any digital infrastructure:
• software for managing grades
• software for registering students
• software for managing courses
=> huge overhead in organizational work
Challenge 3: Infrastructure
Research 47%
Teaching 5%
Co-resp. Mention
Informatique 5%
Other admini-
strational tasks 15%
Responsibility
M2 26%
My work emails sent in 2016
50%
organizational
load
23
Our Proposal
Put the infrastructure in place
BEFORE admitting the students!
But don’t we have Synapses?
(BUR)
An excessive bureaucracy is a real roadblock for an attractive
and efficient working environment.
24
How grades are handled by Synapses
Study
Inspector
Synapses
System
Diploma
System
Head of
SOEIST
school
Lecturer
sends
Excel
returns
Excel
adds course grades
enters grades
manually
Jury of
Master’s
program
computes
final grade
as average
of grades
Google
Sheets
Final
grade
entered
manually
Software
bridge
Head of
Computer
Science
Field
for veri‐
fication
asks to
verify
sends
corrections
co‐
rrects
PDF
List
Printing
verifi‐
cation
sends
corrections
Student
manually prepares
transcript
This is not sustainable!
25
What Synapses does not handle
(Email exchanges about information needs for the Mention Informatique at Paris Saclay in 2017)
26
(Email exchanges about information needs for the Mention Informatique at Paris Saclay in 2017)
The bulk of the administrational
headache is not handled by
Synapses — and will never be!
What Synapses does not handle
27
Plus all the offline forms...
...that we use at the “meilleure école du numérique en France”:
!!!
28
Our Proposal
Provide a Google Sheet that (1) is hosted in France and
(2) allows reading and writing rights for each cell.
(SHEETS)
Example: Collecting lab grades for a course
Each lecturer can
write their column
Each student
can read their line
This is not handled
by Synapses
This is not handled
by Synapses
29
Our Proposal
Provide a Google Sheet that (1) is hosted in France and
(2) allows reading and writing rights for each cell.
(SHEETS)
Example: List of M2 programs
Only the responsible
can update their line
Everyone can
read this information
30
Our Proposal
Provide a Google Sheet that (1) is hosted in France and
(2) allows reading and writing rights for each cell.
(SHEETS)
Example: Research Missions
Each researcher
can read and write
their line
Only the n+1
can give their OK
31
Good news!
There is already a fully functional prototype that everyone can use
with their login from Télécom (by Shibboleth).
(SHEETS)
Thanks to:
• The PAF team
• Julien Romero
• Fabian Suchanek
• Support from DSI
This could be where Télécom ParisTech takes the lead in NewUni!
But: The prototype needs extension, debugging, and maintenance.
32
How to finance the development
(SHEETS)
Email exchanges with
study inspector
Information
request
Grade
update
Synapses
Approve
applicants
Student
following
other
(jury, rooms, housing)
50% are just
updating/
requesting
tabular information
• There was probably money allocated for the infrastructure of NewUni
(we cannot construct a new university without any financial means)
• There was probably money allocated for the grading software
(which Télécom currently does not have but other universities have)
• Otherwise, see what we could save:
33
Challenge 4: Shared Calendar
(CAL)
Current calendar at NewUni:
Problems:
• we cannot share courses between schools
• we cannot share courses between years of study
• administrational headache
Different
for each
year of
study!
Our Proposal
• all courses start in the same week of the year
• all courses end in the same week of the year
• all courses take place in predefined slots of the day
(9:00-12:00, 13:30-16:30)
• the courses have a predefined number of hours per week,
and the type of course + the number of hours defines the ECTS
... no matter the discipline, the year of study, or the department.
9
10
11
12
1
2
3
4
5
6
7
8
9
Mois:
Harvard
Do like all other international universities: Have a shared calendar:
34
(CAL)
Our Proposal
It does not have to be obligatory, it can be a default plan.
What is required:
1) the readiness to support this model (missing in Paris-Saclay)
2) a central authority to propose the calendar (also missing in P-Say)
9
10
11
12
1
2
3
4
5
6
7
8
9
Mois:
Harvard
Do like all other international universities: Have a shared calendar:
35
(CAL)
Challenge 5: Research Organization
36
(COL)
Interdisciplinarity is a goal of NewUni [Conv. de Coop.]
“Si tu veux que tout soit bon, fais une collaboration”
Idea: Ask what the researchers actually think about collaborations.
We believe that collaboration in research is not a goal in itself.
Research does not become better just because it is collaboration.
If we give funding preference to projects that are collaborations,
we pay the opportunity costs of not funding other projects
that are maybe scientifically more reasonable.
Our Proposal:
37
(COL)
Collaboration should not be advertised as a goal, taken
as a measure of success, or prioritized for funding.
NewUni should not nudge people to collaborate with
certain politically desired partners. NewUni should just
help researchers to get to know each other
(through social events, seminars, and team visits),
so that they can collaborate when they see the need.
Challenge 6: Master’s programs
38
Double diplôme IFP School
Double diplôme IFP School
Double diplôme ENSAE ParisTech
Double diplôme avec HEC ParisTech
3e année à Mines ParisTech
12 x 3e année IMT
10 x 3e année ParisTech
Cursus TIC (Montpellier)
MSc Digital HEC
Double diplôme HEC
Double diplôme ENSAE
Double diplôme ENSAE
IDSI (TSP&TÉM)
Formation sur 3 ans
entrée par certains des 30 double‐diplômes internationaux
entrée par M1,
certains des 30 d.d int.,
3e année à l’X
2 de 13 filières
+ ue partagées + 2 ATHENS
+ langues + form. hum.
5 filières
(EURECOM)
5 filières
(EURECOM)
BCI + langues + form. hum.
15 options internes
21 M2 à UPSay
M2 Design Paris1
M2 Biomédical à Paris5
7 M2 à Paris6
M2 Maths à Paris7
M2 Management Paris9
192 FAE (6 mois)
diplôme en VAE
6 Mastères Spécialisés
14 formations CES
3 Executive Mastères
1e année
2e année
3e année
4e année
entrée par L3, DUT, classe prépa,
certains des 30 d.d int.
Formation continue
Corps des Mines
Validation des Acquis
de l'Expérience
Retour à l’établissement d’origine
192 FAE (6 mois)
Départ en double-diplome international
Départ en double-diplome international
Double diplôme HEC
Télécom ParisTech has more possible double‐double diplomas
than students in the third year! (300)
Challenge 6: Master’s programs
39
Double diplôme IFP School
Double diplôme IFP School
Double diplôme ENSAE ParisTech
Double diplôme avec HEC ParisTech
3e année à Mines ParisTech
12 x 3e année IMT
10 x 3e année ParisTech
Cursus TIC (Montpellier)
MSc Digital HEC
Double diplôme HEC
Double diplôme ENSAE
Double diplôme ENSAE
IDSI (TSP&TÉM)
Formation sur 3 ans
entrée par certains des 30 double‐diplômes internationaux
entrée par M1,
certains des 30 d.d int.,
3e année à l’X
2 de 13 filières
+ ue partagées + 2 ATHENS
+ langues + form. hum.
5 filières
(EURECOM)
5 filières
(EURECOM)
BCI + langues + form. hum.
15 options internes
21 M2 à UPSay
M2 Design Paris1
M2 Biomédical à Paris5
7 M2 à Paris6
M2 Maths à Paris7
M2 Management Paris9
192 FAE (6 mois)
diplôme en VAE
6 Mastères Spécialisés
14 formations CES
3 Executive Mastères
1e année
2e année
3e année
4e année
entrée par L3, DUT, classe prépa,
certains des 30 d.d int.
Formation continue
Corps des Mines
Validation des Acquis
de l'Expérience
Retour à l’établissement d’origine
192 FAE (6 mois)
Départ en double-diplome international
Départ en double-diplome international
Double diplôme HEC
+ Option IA
+ Graduate Degree IA
+ Graduate Degree Cyber
+ MS Spécialisé
IA
This is not how the MIT looks!
Challenge 6: Master’s programs
40
Double diplôme IFP School
Double diplôme ENSAE ParisTech
Double diplôme avec HEC ParisTech
3e année à Mines ParisTech
12 x 3e année IMT
10 x 3e année ParisTech
Cursus TIC (Montpellier)
MSc Digital HEC
Double diplôme HEC
Double diplôme ENSAE
IDSI (TSP&TÉM)
Formation sur 3 ans
entrée par certains des 30 double‐diplômes internationaux
entrée par M1,
certains des 30 d.d int.,
3e année à l’X
2 de 13 filières
+ ue partagées + 2 ATHENS
+ langues + form. hum.
5 filières
(EURECOM)
5 filières
(EURECOM)
BCI + langues + form. hum.
15 options internes
21 M2 à UPSay
M2 Design Paris1
M2 Biomédical à Paris5
7 M2 à Paris6
M2 Maths à Paris7
M2 Management Paris9
192 FAE (6 mois)
diplôme en VAE
6 Mastères Spécialisés
14 formations CES
3 Executive Mastères
1e année
2e année
3e année
entrée par L3, DUT, classe prépa,
certains des 30 d.d int.
Formation continue
Corps des Mines
Validation des Acquis
de l'Expérience
192 FAE (6 mois)
Départ en double-diplome international
+ Option IA
+ Graduate Degree IA
+ Graduate Degree Cyber
+ MS Spécialisé
IA
Problems:
• administrational overhead
• no deep specialization
• no room for small fields
• non‐validation of a course means repeating a year
(25% of emails for M2 responsibility=time not spent with students)
(M2 = 7 weeks introduction, 7 weeks specialization, then internship)
(Cannot have a full M2 on Human Computer Interaction)
Proposal: 2 year shared master’s
41
(MERGEMASTER, 2YM; not voted by DIG)
School A
2nd year
(M1)
3rd year
(M2)
External
admission
1st year
Master’s Program
School B
Advantages:
• full compatibility with Bologna
• simplicity
• flexibility (variable part in Master)
• more choice for the students
• attractive master’s programs
Schools can
keep their
students 2 days
a week if
they want.
Engineering diploma
is different from
master’s diploma,
if desired
Proposal: No parcours, just fields
42
(FIELDS, not voted by DIG)
Internship / Research Project
Artificial
Intelligence
Networks
HCI
Cultural
training
Software/
Systems
Data
Sciences
Security
English/
French
You validate the master’s program, if you validate
• 60 ECTS in a topic of your choice (“major”)
• 20 ECTS in other topics of your choice (“minors”)
• 10 ECTS in “cultural training”
• 30 ECTS in the internship
Master’s
program
(2 years)
Pool of
courses,
some with
prerequisites
PhD
Bachelor
level
(3 years)
Advantages:
• flexibility
• simplicity
• choice
• easy administration
• existence of small fields (HCI)
• like
the MIT
and
others