SIGCSE

Announcements

• Talking “When Are We Gonna Have a Serious Conversation About How Unserious (and Problematic) SIGCSE Is?”
• We will sigcSEE how the conversation goes?
  • How many minutes until spring break.
• Homework
  • See my spring break comment.

Review Question 1

Computing & Data Programs offers a Computer Science degree that achieves coverage of the "Computer Science Curricula 2013" Core Tier-1 topics targeting 165 total hours of lecture instruction. How many required courses within the major were needed to meet this requirement?

Review Question 2

Computer Science degrees are recommended to cover at least 90% of Core Tier-2 topics of an additional 143 total hours of lecture instruction. How many required courses within the major were needed to meet both these requirement if 293.7 hours?

Review Question 3

Computing & Data Programs offers a Data Science degree for which no national accepted curricular standards yet exist. How many hours of lecture are met by major-required courses in Data Science?

Review Question 4

Topics with more Tier-2 hours than Tier-1 hours includ "Networking and Communication", "Software Engineering", "Intelligent Systems", and "Information Management". How are these topics organized at Willamette?

Review Question 5

"Discrete Structures" and "Software Development Fundamentals" require 41 and 43 hours respectively, and are not included in Data Science requirements but are Tier-1 topics. Depending on elective selection, could a Willamette Data Scientist be considered to have completed the requirements of an undergraduate Computer Science degree?

~Review Question 6

"Social Issues and Professional Practice" have 11+5 hours, and "Programming Languages" have 8+20 hours. With the 90% rate for Tier-2 Core, this gives 41.5 hours total. How many hours of instruction in Data Ethics do you think will focus on HTML/CSS/JavaScript, a universal "second language" to CS-151 and DATA-151 students?

~Review Question 7

Do you believe computer science education can and should be approached scientifically.

Peer Review

Peer review is currently an expected part of the scientific process. In CS:

• We have workshops, conferences, and journals, which publish papers.
• For each, there are program committees that would to accept or reject.
• These program committees ususally assign 2-3 reviewers to each submission.

In CS, there are some limitations:

• Most program committees are uncompensated
• Research is increasingly specialized the point reviewers are rarely qualified.
• Reviewers are often chosen from established academics, and have an incentive to accept submissions that cite their own work.

Bias in review is nominally addressed by single or double-masking.

Bias Case Study

I have had three advisors: Prof. Andrew Chien (B.S.), Prof. Sanjoy Baruah (M.S.), Prof. Cynthia Sturton (Ph.D.).

• I achieved a 100% publication rate with Prof. Chien and Prof. Baruah.
• My first paper with Prof. Sturton was rejected, all others were accepted.

Prof. Sturton's Hypothesis: Our publication was rejected because the title suggested it was authored by women even when author names were redacted.

Bias Case Study

In CS, many publications have titles draw from the arts and humanities. Here are the first three I found for which I am not a co-author:

• AKER: A Design and Verification Framework for Safe and Secure SoC Access Control (HOST)
• CS-JEDI: Required DEI Education, by CS PhD Students, for CS PhD Students (SIGCSE)
• Javert: Fully automatic mining of general temporal properties from dynamic traces (SIGSOFT)

Bias Case Study

I do not claim to have executed sampling correctly, but all three of these names are or slant masculine.

• Aker is a Egyptian deity of the underworld, depicted as a lion with a mane.
• All Jedi were men from 1978 to 1994, and comfortable majority of named Jedi of known gender (200/334) are men.
• Javert is man in Les Misérables

Bias Case Study

Rui Zhang, myself, Peng Huang, and Cynthia Sturton submitted "Coppélia: End-to-End Automated Exploit Generation for Validating the Security of Processor Designs"

• Coppélia is woman in a ballet.

Our paper was rejected twice under this title, then accepted and won a best paper award when removing "Coppélia" from the title.

Bias Case Study

Are peer reviewers evaluating quality? Or are they neural networks computing a gender bit based on what they imagine when reading titles?

Crisis

In science we have a thing called the "Replication crisis"

• Review quality is low, and reviewers are overwhelmed
• Review work is uncompensated.
• Professor/research real wages are falling while labor demands increase.
• There are fewer jobs per Ph.D. than any point in history

These may be manageable but philosophers of science are beginning to recommend alternative models.

• Read more: Is Peer Review a Good Idea?

Personally, I believe peer review is an impediment to scientific progress.

Read More

I have both written and received some peer reviews, which I believe are appropriate to share.

• Some I wrote
• Some I got

In the case of the ones I got, you can read the published works. In other cases, they are not yet available.

• My writing is here.

We will now take n minutes to review reviews.

Discussion

Discussion Time.

• Read “When Are We Gonna Have a Serious Conversation About How Unserious (and Problematic) SIGCSE Is?”