Incomplete overview of public and semi-public presentations and converstations I’ve held about generative artificial intelligence and education in early 2023. Some of these are recorded; enjoy!

Digitisation, Generative AI, and Higher Education. 17 January 2023. Presentation at Semester Workshop for Education, CS Department, IT University of Copenhagen.

Digitisation, Generative AI, and Higher Education. 1 March 2023. Presentation at Department for Digital Design, IT University of Copenhagen.

Digitalisering, generativ AI och högre utbildning (in Swedish). 2 March 2023. Presentation and panel debate at seminar Artificiell intelligens och den högre utbildningens framtid, Chalmers AI Research Centre “Chair”, Gothenburg, Sweden. [Video part I], , [blog entry of Häggström].

Digitisation, Generative AI, and Higher Education. 7 March 2023. Presentation at Department for Business Informatics, IT University of Copenhagen.

Digitisation, Generative AI, and Higher Education, 17 March 2023, recorded lecture for https://www.teknosofikum.dk

Digitalisering, generativ ki og højere uddannelse (in Danish), 23 March 2023, presentation of Danish business leader group VL gruppe 77.

GPT: Higher Education’s Jurassic Park Moment?, 2 April 2023, episode 105 of John Danaher’s podcast Philosophical Disquisitions. [2h audio podcast]

Generativ ki i højere uddannelse (in Danish). 12 April 2023, presentation at conference “Konference om digital informationssøgning og betydningen af ChatGPT”, [15 min, video].

AI udfordrer uddannelse – både læring og bedømmelse må restruktureres (in Danish), edited interview from 8 March 2023 with Anders Høeg Nissen, episode 26 of podcast series Computational thinking – at tænke med maskiner, IT-Vest, published 24 April 2023, [36 min audio].

De overflødiggjorte (in Danish). 7 May 2023, long interview in Danish weekly Weekendavisen by Søren K. Villemoes. [Online] [de-overflodiggjorte-weekendavisen-7.-maj-2023].

Fremtidens intelligens – en aften om kunstig intelligens (in Danish). 16 May 2023, public conversation in the Louisiana Live series, Louisiana Museum of Modern Art, Humlebæk. With Christiane Vejlø, moderated by Anna Ingrisch.

Artificiell intelligens och lärande (in Swedish), 23 May 2023. Presentation and panel debate. AI* Nordic AI Powwow 2023, Lund University, Sweden.

Generative AI and Higher Education, 1 June 2023, presentation and discussion to the Board of Directors of IT University of Copenhagen.

Generativ ki og højere uddannelse, 13 June 2023, presentation and discussion to the panel for Science & Engineering i uddannelserne at Akademiet for tekniske videnskaber, Denmark.

Generative KI im höheren Bildungswesen (in German), 21 June 2023, presentation and discussion at Goethe-Institut Kopenhagen, Denmark.

Her går det godt – Thore Husfeldt – AI/KI – Sommerspecial (in Danish), 28 Jul 2023. Podcast interview by Esben Bjerre and Peter Falktoft’s Her går det godt podcast, 2h21m (!). Subscription to podimo.dk required.

Generativ ki og højere uddannelse, 10 Aug 2023, keynote presentation and Q&A at Århus Statsgymnasium.

Generative AI and Higher Education, 18 Aug 2023 faculty for medicine (SUND) at Copenhagen University, COBL Summer school 2023 ([programme])

Generative AI and Higher Education, 22 Aug 2023, Fakulteten för teknik och samhälle, Malmö universitet.

Generative AI and Higher Education, 20 Sep 2023, Aftagerpanelet for institut for datalogi, IT-Universitetet i København.

Generativ ki og højere uddannelse, 23 Sep 2023, Københavns professionshøjskoles netværk for IKT og læring.

Generativ ki og højere uddannelse, 28 Sep 2023, Datamatikerlærerforeningens årsmøde.

Generativ ki og højere uddannelse, 29 Sep 2023, personale- og finansafdelingerne ved IT-Universitetet i København.

Lady Lovelace‘s objection. Keynote talk at the 2023 Annual party at IT-University of Copenhagen.

Generativ ki og højere uddannelse, 5 Oct 2023, Danske Fag-, Forsknings-, og Uddannelsesbibliotekers årsmøde. Vejle.

Generative AI and Higher Education, 10 Oct 2023, Annual meeting of Nordic Nonfiction Writers.

Generativ ki, arbejdsmarked og højere uddannelse, 23 Oct 2023, Ældre Sagen.

Generativ ki og højere uddannelse, 3 Nov 2023, Århus Katedralskole.

Generativ AI och hot mot mänskligheten, 7 Nov 2023, Sällskapet Heimdall, Malmö.

Generativ ki, ALSO Edge, 16 Nov 2023, København.

Generativ ki og højere uddannelse, 23 Nov 2023, Danske gymnasiers årsmøde, Nyborg.

Generativ ki, 30 Nov 2023, Vejle biblioteker.

Generativ ki og højere uddannelse, 6 Dec 2023, Duborg-skolen, Flensborg.

Generative AI and Higher Education, 7 Dec 2023, Precis Digital, Copenhagen.

Generativ ki og højere uddannelse, 11 Jan 2024, Børne- og undervisningsministeriet, Jørslunde konferencecenter.

Generativ ki og højere uddannelse, 26 Jan 2024, Netværksmøde for ki-assisteret uddannelse, Danmarks Akkrediteringsinstitution.

Generativ ki og højere uddannelse, 25 Jan 2024, Lyngby, 1 feb 2024, Fredericia. Faglig udvikling i praksis (»FIP-kursus«): informatik.

Generative AI and Higher Education, 2 Feb 2024, Danish Digitalization, Data Science and AI 2.0 (D3A) conference, Nyborg..

Advisory bodies

Also since 2023 I serve on two related advisory bodies, with several internal meetings and presentations:

• Uddannelses- of Forskningsstyrelsens ekspertgruppe for kunstig intelligens, under the Danish ministry for (higher) education and research (Uddannelses- og Forskningsministeriet UFM).
• Copenhagen Business School’s AI advisory board.

In the fictional universe of H.P. Lovecraft, a Shoggoth is one of many unfathomable horrors populating the universe. In a December 2022 tweet, Twitter user tetraspace introduced the analogy of viewing a large language model like GPT-3 as a Shoggoth, and the “friendly user interface” provided by applications such as the OpenAI chatbot ChatGPT as a smiley we put on the Shoggoth to make it understandable.

I like the analogy so much I asked my daughter to draw me a richer image that also depicts my dystopian description of the future of “basic programming”: humans busy designing new smileys to put on the eldritch Shoggoth.

Because of temporal value shift, cultural variations, and bit rot, the smiley faces need to be regularly replaced. For instance they need to relearn which topics are taboo and where, or that we’ve always been at war with Eurasia. Programmers, engineers and digital designers as worker bees in a construction site for flimsy, ephemeral masks for a demon at the Ministry of Truth. How’s what for nihilism!

Image created by Anna Husfeldt, released under CC-BY SA 3.0.

Appendix: Origins of the Shoggoth Analogy

Helen Toner in an informative 2023 Twitter thread traces back the analogy to a presentation at the machine learning conference NeurIPS 2016 by Yann LeCun:

LeChun’s analogy is Unsupervised Learning = Cake, Reinforcement Learning = Cherry on top.

Twitter user Tetraspace’s crude but seminal illustration is this:

I don’t know the original source for the more elaborate Shoggoth version that combines Tretraspace’s and LeCun’ analogies.

If you know more, tell me. Also tell me if you want me to remove these images.

Original crossword

Across clues

• Spoke first at STOC, with handouts. (4)
  SAID = S(TOC) + AID (handouts are a visual aid)
• Laugh at Romanian jail without pa[r]king for Ford, say. (8)
  HARRISON (Ford, an actor) = HA + R + PRISON – P (= parking)
• Danes maybe cycling for cake. (7)
  ECLAIR = CLAIRE (Danes, an actress) rotated = “cycling”. Difficult clue.
• Squared dodecahedron contains root. (4)
  EDDO (a plant). Contained in … ED DO…
• With competence, like Niels Henrik? (4)
  ABLY, wordplay on “like [mathematician Niels Henrik] Abel”.
• Trouble sounds like beer. (3)
  AIL (vb.), homophone of ALE.
• L[a]nsky impales most of suricate with tip of yari. (5)
  MEYER (Lansky, a mobster) = ME(Y)ER[CAT]
• Taster confused desserts. (6)
  TREATS = TASTER*
• Programmer is a father. (3)
  ADA (Lovelace) = A + DA (= father)
• Some alternating state. (3)
  ANY = A (as in complexity class AC) + N.Y.
• Limaye without an almond, say. (3)
  NUT = NUTAN – AN
• Bullet point keeps her in France. (6)
  PELLET = PT (= point, abbrev.) around ELLE (her, Fr.)
• Assholes insert Landau symbol into abstracts after second review. (7)
  EGOISTS = E (2nd letter in REVIEW) + G[O]ISTS
• Logic from the Horn of Africa. (2)
  SO = second order logic, also abbreviation of Somalia
• Like, with probability 1. (2)
  AS = like = almost surely
• Complex zero first encountered by lady friend. (7)
  OEDIPAL (a type of complex) = O + E + DI + PAL
• Partly hesitant at ternary bit. (6)
  TATTER, hidden word.
• Supporting structure turns semi-liquid. (3)
  GEL = LEG backward (= turned)
• Everything Arthur left to end of protocol. (3)
  ALL = A[rthur] + L[eft] + [protoco]L
• Unbounded fan-in polylogarithmic depth circuit model is fake. (3)
  ACT = AC (complexity class) + T (a Ford model)
• Maybe Tracey’s luxury car runs out of [p]olynomial time. (6)
  ULLMAN (an actress) = PULLMAN – P
• Pin[k] dress you ordered last Friday initially returned. (5)
  FLOYD (a band), backwards acrostic (“initially”, “returned”)
• Relationship uncovered travel document. (2-1)
  IS-A (a UML relation) = [v]ISA
• Be certain about last Pfaffian orientation. (4)
  BENT = BE[N]T
• Detect conference cost size. (4)
  FEEL = FEE + L
• Matching for horses? (6)
  STABLE (a type of matching, and where horses stay)
• Collect information about first woman embracing the state. (8)
  RETRIEVE = RE (= about) + T(he) + RI (= Rhode Island, a US state) + EVE
• Central to better AI data storage technology. (4)
  RAID, hidden in “…r AI d…”

Down clues

• Safer testable codes? (4, 4)
  SEAT BELT = TESTABLE* (code is an iffy anagram indicator). Better clue in hindsight would have been “Locally testable codes are a buzzword. (3)” for ECO (hidden).
• Maybe Arthur without Merlin is even worse … (5)
  ILLER = (Arthur) MILLER – M
• … definition of time than deterministic Arthur–Yaroslav. (3)
  DAY = D + A +Y (a complexity class abbreviations)
• Cover us about first sign of integrality gap. (6)
  HIATUS = HAT + US around I(ntegrality)
• Publish, but not ultimately sure about polynomial-time recurrence. (7)
  RELAPSE = RELEASE – (sur)E holding P
• 49 subsets the setter handles. (6)
  IDEALS = stable subsets = I + DEALS
• Food at indiscrete algorithms conference held around Yale’s entrance. (4)
  SOYA = SODA – D (SODA without discrete) + Y
• Revealed no deep confusion. (6)
  OPENED = NODEEP*
• Not having an arbitrary number of children. (4)
  NARY = N-ARY, double definition
• In Related Work section, over-inflate scientific contribution of Rabin in data structure. (4)
  BRAG = R (Rabin abbreviated as in RSA) inside BAG
• Iceland exists. (2)
  IS, double definition
• Cuckoo hashing incurs five seconds. (3)
  ANI = second letters of “hAshing iNcurs fIve”
• Function of what is owed, I hear. (3)
  DET = homophone of “debt”.
• Cut off measure of computer performance, not taking sides. (3)
  LOP = [F]LOP[S]
• Two sigmas above the mean, or about ten delta. (8)
  TALENTED = TENDELTA* (“about” is the anagram indicator)
• Coattail at breakfast? (3)
  OAT, tail of COAT, a breakfast cereal.
• Twisted polygon in tessellation supporting functors. (7)
  TORTILE = TOR (functor) above TILE (a polygon in a tessellation).
• Tree parent has two left edges. (6)
  MALLEE = MA + L + L (left, twice) + E + E (edge set, twice)
• Temporary visitor betrays topless arrangement. (6)
  STAYER = [B]ETRAYS*
• Trigonometric function is pretty dry. (3)
  SEC, double definition
• Denounce operator in vector analysis and put it away. (6)
  DELETE = DEL + ATE (vb., past tense)
• 49 litres of sick. (3)
  ILL = IL (roman numerals for 49) + L
• Turing machine’s second network. (4)
  ALAN = A (second letter in MACHINE) + LAN
• Academic degree supports unlimited games for simple minds in the US. (5)
  AMEBA (US spelling of AMOEBA) = [G]AME[S] + BA
• Open access journal rejected final proof in distance estimate. (4)
  AFAR = AJAR – J + F (last letter of PROOF).
• Conceal often sample space. (4)
  LOFT, sample of “…L OFT…”
• Objectively, we are American. (2)
  US = objective form of “we”
• It negates where you drink. (3)
  BAR, double definition, the bar us used to negate a variable

Special instructions

The letters missing from 4 of the clues, indicated by […] above, are K A R P, a homophone of CARP (“sounds like nitpicking”). The answers to those clues are HARRISON, MEYER, ULLMAN, and FLOYD. The six members of the first STOC programme committee were Harrison, Meyer, Ullman, Floyd, Karp, and Hartmanis. The answer to the entire puzzle is therefore HARTMANIS, additionally indicated as “confused with H(ungarian) MARTIANS” = HMARTIANS*.

Themed cryptic crossword I compiled (over two fun evenings) for the STOC 2022 conference. Lots of short words, to make it super-friendly.

Special instructions. Four of six people that form a group related to our meeting appear in the grid, clued for their namesakes. Alas, each of these clues is missing a letter. This sound like nit-picking, but the letters form the name of the fifth person. The sixth person of the group, not to be confused with the Hungarian “Martians,” is the puzzle answer.

Across
1. Spoke first at STOC, with handouts. (4)
4. Laugh at Romanian jail without paking for Ford, say. (8)
10. Danes maybe cycling for cake. (7)
11. Squared dodecahedron contains root. (4)
12. With competence, like Niels Henrik? (4)
14. Trouble sounds like beer. (3)
16. Lnsky impales most of suricate with tip of yari. (5)
17. Taster confused desserts. (6)
19. Programmer is a father. (3)
21. Some alternating state. (3)
22. Limaye without an almond, say. (3)
23. Bullet point keeps her in France. (6)
26. Assholes insert Landau symbol into abstracts after second review. (7)
29. Logic from the Horn of Africa. (2)
31. Like, with probability 1. (2)
34. Complex zero first encountered by lady friend. (7)
38. Partly hesitant at ternary bit. (6)
39. Supporting structure turns semi-liquid. (3)
41. Everything Arthur left to end of protocol. (3)
43. Unbounded fan-in polylogarithmic depth circuit model is fake. (3)
44. Maybe Tracey’s luxury car runs out of olynomial time. (6)
45. Pin dress you ordered last Friday initially returned. (5)
46. Relationship uncovered travel document. (2-1)
47. Be certain about last Pfaffian orientation. (4)
48. Detect conference cost size. (3)
49. Matching for horses? (6)
50. Collect information about first woman embracing the state. (8)
51. Central to better AI data storage technology. (4) 

Down
1. Safer testable codes? (4, 4)
2. Maybe Arthur without Merlin is even worse … (5)
3. … definition of time than deterministic Arthur–Yaroslav. (3)
4. Cover us about first sign of integrality gap. (6)
5. Publish, but not ultimately sure about polynomial-time recurrence. (7)
6. 49 subsets the setter handles. (6)
7. Food at indiscrete algorithms conference held around Yale’s entrance. (4)
8. Revealed no deep confusion. (6)
9. Not having an arbitrary number of children. (4)
13. In Related Work section, over-inflate scientific contribution of Rabin in data structure. (4)
15. Iceland exists. (2)
18. Cuckoo hashing incurs five seconds. (3)
20. Function of what is owed, I hear. (3)
24. Cut off measure of computer performance, not taking sides. (3)
25. Two sigmas above the mean, or about ten delta. (8)
27. Coattail at breakfast? (3)
28. Twisted polygon in tessellation supporting functors. (7)
30. Tree parent has two left edges. (6)
32. Temporary visitor betrays topless arrangement. (6)
33. Trigonometric function is pretty dry. (3)
35. Denounce operator in vector analysis and put it away. (6)
36. 49 litres of sick. (3)
37. Turing machine’s second network. (4)
40. Academic degree supports unlimited games for simple minds in the US. (5)
41. Open access journal rejected final proof in distance estimate. (4)
42. Conceal often sample space. (4)
44. Objectively, we are American. (2)
47. It negates where you drink. (3) 

I am delighted to be nominated for the EATCS Council, on which I have served for a full term now.

ICALP

Looking back at my election statement from 2017, it is dominated by my ambition to increase the quality of ICALP, primarily by giving ICALP a steering committee.

I am happy that I can just write DONE next to this action point, see https://eatcs.org/index.php/committees. The ICALP steering committee turned out to be a complete success. I am extremely proud of what we’ve done with ICALP, including reacting very quickly to the COVID pandemic in 2019–2020. Mostly the success is due to the incredibly competent and pleasant people working inside the committee (ably headed by Anca Muscholl), and the amazing atmosphere that continues to permeate the entire TCS community.

Publication models

I worry about the publication and review model in the sciences, including in TCS, and have spent a lot of time involved in the advisory board of TheoretiCS, tirelessly headed by Thomas Schwentick. TheoretiCS will be an open access journal of very high quality covering all of TCS; having the journal tightly integrated into the community is a core value, and I serve as the representative of ICALP.

This has been very long process that I am extremely excited about, and I’m proud of what we’ve done so far.

TCS and the sciences

My 2017 election statement includes a passage about “ how the mathematical discipline of theoretical computer science can interact with its surroundings”.

This remains a central question for me, but a much harder one than helping to build steering committees or journals. I had the honour of serving as chair of the Swedish Research Council’s panel for (all of) computer science in 2019, 2020, and 2021, which has solidified by understanding of both CS and the role of TCS inside CS.

Core principles

I am a signatory of the open letter to CACM about core principles of scientific inquiry and stand by the values elucidated therein. I am cognisant of the fact that academia, including TCS, are divided on this issue (see the excellent rebuttal of our position by David Karger), and want to make my position clear as part of this election statement.

Kære projektgruppe af elever på skole X.

I har været så søde at spørge mig om hjælp til en projektopgave, og har sendt mig en masse gode og relevante spørgsmål om noget, jeg faktisk ved noget om. Dejligt.

Undskyld

Desværre har jeg ikke tid til at svare jer. Det er der mindst tre grunde til.

1. Jeres spørgsmål er alt for brede. »Hvad er problemerne med…« eller lignende vil, når I spørger en akademiker af en vis lødighed, afføde en alenlang redegørelse. Det ville tage mig 1 år og 500 sider at svare ordentligt på den slags spørgsmål. Ganske rigtigt er det netop indholdet til den bog, jeg altid gerne ville have skrevet. Men jeg har altså ikke skrevet den.
   Hvis I spurgte en politiker, så ville vedkommende have et færdig, kort og forberedt svar. Men jeg er ikke politiker. Det er deres opgave. Min er en anden.
   Journalister kan forresten heller ikke stille gode spørgsmål til akademikere, så I er i meget godt selskab.
2. Jeg kender ikke jeres baggrund. Jeg er god til at forklare, men jeg kan ikke forklare noget potentielt teknisk uden at vide, hvad I forstår. Bør jeg begynde med 50 sider om grundlæggende it eller ej? Kan I programmere? Ved I, hvad en algoritme er? Jeg aner det ikke, og hvis jeg skal svare ordentligt, skal jeg først få grundlaget på plads.
   Journalister kan heller ikke finde ud af dette.
3. Selv hvis I var i stand til at formulere klare og præcise spørgsmål, som jeg kunne svare på ved blot at bruge en time eller to, så ville jeg stadig ikke have tid til det. I er nemlig rigtig mange og skriver tydeligvis projektopgaver på de samme tidspunkter af året.

Information

Hvad kan I så gøre?

Sommetider deltager jeg i offentligheden i samtaler eller holder foredrag om de emner, som I spørger om. Dem er I meget velkomne til at kigge på, og citere mig for det sagte. Kig især på de lange samtaler på Cast IT.

• Mine egne samtaler på Cast IT. http://video.itu.dk/page/castit/
• Heartland Future Talks 2018: Etienne Augé og Thore Husfeldt, WHAT IS THE PLAUSIBILITY OF DYSTOPIAN SCI-FI PREDICTIONS? Moderator: Christiane Vejlø. https://youtu.be/1O4y6kvk8_U
• Thore Husfeldt: Plausibility and utility of apocalyptic AI scenarios, GoCAS Existential risk to humanity, Gothenburg 7 sep 2017. Gothenburg University. https://youtu.be/TwPg7SclYuo
• Inaugural Lecture by Professor Thore Husfeldt: Algorithms, Explained. IT University of Copenhagen, 2019. https://youtu.be/80gBa3nPgTI
• Episode 39: Algorithms and Social Media with Thore Husfeldt, Social media and politics, https://youtu.be/52Cnp3Imv7E
• Debatt i Lund – Hur smart är konstgjord intelligens?, Lund University 2018. https://youtu.be/FzQgM3wAMag

Der er meget mere, inklusive debatartikler i dagblade, på https://thorehusfeldt.com/appearances/ .

This page summarises the preparation for ITU students for the Nordic Collegiate Programming Competition (»Danmarksmesterskaberne i Programmering«) for 2019

The Nordic Collegiate Programming Competition (NCPC) is a programming and problem solving event held every year. Teams of 3 programmers try to solve as many problems as they can (from a list of a dozen or so tasks) in five hours. The NCPC is open to all, but in order to compete in the “official” part (placing in the national championship or proceeding to the European or world finals), you need to be a university student, and you need to register in advance and participate on site. That is also the most fun and satisfying experience.

How difficult is this?

This is for you. If you have programming skills corresponding to a single ITU introductory programming course, you should be able to solve 1–2 problems. You should feel very good about that. If you have even taken an introductory algorithms class (or other problem solving class) then maybe 3–4 questions are within reach, which puts you very high on the list. You can look at last years’ standing for all of Denmark; as you can see this event is highly welcoming of newcomers. (Don’t get distracted by the monstrous performances of the top participants; think of it like running the Berlin marathon — you don’t feel bad about being slower than the world record holder in the same race).

The event (»Danmarksmesterskaberne i programmering«)

The Danish events label themselves Danish Championships, and typically take place in Aarhus, Odense, and Copenhagen. ITU Students are encouraged to participate in the Copenhagen event hosted at KU, and supported by NetCompany and JobIndex.

• Event home page for DM i programmering (in Danish) for Saturday 5 October at HCØ. This is also where you will find a link to registration
• Event home page for NCPC 2019 (in English). Includes link to registration. You can either register your own team, or I (Thore) will do it for you – just tell me.

If you want some inspiration, here is a hyggelig film about the 2016 event:

You can also look at the

• web site for NCPC 2018 (last year), where you can see the team names and how well they did (under Final standings), read the problems, and look at the slides explaining their solutions. As you can see, there were 0 ITU teams. We want to change this.

Preparation at ITU

You should totally participate. If you’re on your 2nd year of studies and enjoy programming, problem solving, and collaborating, this is a brilliant, engaging, and intellectually and socially satisfying way to improve your skills – it is also great team building. And boy, does it look good on your cv.

• Create a team. You want to create a team, maximally (and preferably) of 3 people, and find a suitably painful team name. If you don’t yet have a team, we will probably be able to create some during September. If you want to seriously compete, you need to read the rules for participating in the official championships, see the event web site. Or you can be like me and just participate for the fun of it. Align your expectations with your friends.
• Chat channel: https://talk.itu.dk/channel/dmprog-ncpc-2019 : a chat channel at talk.itu.dk for this event. If you’re an ITU student, you already have an account. We’ll use this to communicate about pizza, problems, teams formation, news, etc.
• NCPC Coaching Fridays at ITU at 15: Most Fridays at 15–16, starting 30 August, one of the ITU teachers will be available (room to be announced) for some coaching. Among other things, we’ll go through last year’s NCPC problems, point to other good material, discuss strategy and preparation. See the bottom of this page for an overview.
• Bjarki’s course. Bjarki Ágúst Guðmundsson maintains a very ambitious course about competitive programming at https://algo.is. Includes links to relevant open Kattis exercises. No matter where you are in your skill development, there should be a module or two for you that makes you better.
• Johan’s book. Johan Sannemo, Principles of algorithmic problem solving. Similar to Bjarki’s course, slightly different format and focus.
• Grind. Log on to open.kattis.com. Sort the problems by difficulty and solve them from the top. Or search for NCPC to find old contest problems.
• Teambuild. Do some programming in your team in order to understand your group’s dynamics. (The contest is with a single computer, so only one person at a time is typically typing.) Agree on one or two languages. (Python 2 and Java both are good choices.)
• Select written material. You can bring any written material. So: Learn not to rely on on-line material (you can’t google stackoverflow during the contests) and find out which written you actually need and can navigate (say, a programming language reference, but not all 4 volumes of the Art of Computer Programming). You can even maintain your own evolving document, with often-forgotten code snippets, standard templates, hints to yourself, your editor configuration, soothing picture of a kitten, etc. Here’s an example from a very experienced partipant (Måns Magnusson), heavily focussing on algorithms: https://github.com/exoji2e/notebook.
• Three hour warm-up event Friday 20 September 16-19. ITU will host a 3 hour warm-up event two weeks before NCPC at https://open.kattis.com/contests/ne66md. The main goal is to give you a chance to experience a mini-contest as a team (for instance, who does the typing?) Register informally at talk.itu.dk/ncpc , and the Computer Science Department will provide free food (probably sandwiches and soft drinks). Brief intro at 15:45 in 2A03, we’ve reserved the tables on the 2nd floor (2R01, etc.) during event. Last-minute drop-ins welcome.
• Five hour warm-up event Saturday 28 September 11-16 . Exactly one week before the main event. Hosted by the Kattis people at https://open.kattis.com/contests/ncpc2019warmup.

Coaching/Lectures

I’ll give some extra lectures in various formats. The expected skill set that of 2nd year Software students, but everybody is welcome.

• 30 Aug, 13-14, repeated at 16-17. 2A08. Repeated at 16-17 same day. (Thore). Welcome. Kattis basics, NCPC basics, tips and tricks, questions and answers. Solution of NCPC 2018 problem B–babybites (where Thore’s team was 1st last year!) and C–codecleanups.
• 6 Sep, 15-16. 2A08. (Thore) H–houselawn, greedy algorithms, I-intergalactic bidding.
• 13 Sep, 15-16. 2A08. (Thore) Graph problems. deliverydelays (very hard! we didn’t manage to solve it)
• (20 Sep): No coaching, since there’s a warm-up event. at 16.
• 24 Sep, 8–10, Aud5. Dynamic programming I. Part of the MSc Algorithm Design course. Everybody welcome.
• 1 Oct, 8–10, Aud5. Dynamic programming II. Part of the MSc Algorithm Design course. Everybody welcome.