Blog, ECS 188, Winter 2022

Remember, you are required to read the blog at least once per day. Usually I will also send an e-mail note announcing that I've made a new blog post.

I just submitted my ECS 188 course grades to the Registrar. You can compute your grade from the formula in the class syllabus, but if you wish for confirmation, feel free to e-mail me.

About 30% of you received some form of A grade. BTW, the formula does not cover A+ grades; I did give two of them. Most of the rest of the grades were some form of B.

BTW, I mentioned in class that the CS Dept. is quite concerned about alarming grade inflation that has arisen in the last, say, five years. I spoke to you in class about this, and consider it a very serious problem. The department will likely institute serious reform in the coming weeks. However, it is recognized that a discussion course like ECS 188 is different, and that most instructors will grade more liberally in this course.

Separately, I will be sending you one more e-mail message (not on the blog), hopefully later this afternoon. Please read it--you may be surprised at its contents!

Congratulations on finishing what many found to be a long, trying quarter.

Very sorry for the delay in setting your course grades. I now have all the records collected and set up to run through my grading script, which I will do tomorrow. For now, though, I will be sending you the records I have for you. Note that they will be in the format

e-mail, exam 1, exam 2, participation, name, presentation, random Qs

The policies on course grades are delineated in our course syllabus, from which you should see in advance what your grade will be. The formula will give you a lower bound, which in most cases will be your actual grade. I do sometimes bump up a grade produced by the formula for my course. I do also have a proviso for cases in which a student did very little, which may affect one or two grades here.

I'm sending out the Exam 2 grades now. Overall they were very good. One common problem was that often a student would supply a question form our readings, but one that did not address the specific issue raised by the prompt. Also, a couple of students apparently did not know the meaning of the word "passage," which I had not anticipated.

Answers are available in the OldExams directory in our class Web site.

Exam 2 will be held tomorrow in the lecture time slot, 12:10-1. Make sure to have OMSI on your computer, and make sure you can use it to access campus entities. (Run your own server on CSIF as a test run.) You may freely access the Internet, but are not allowed to communicate with others in any manner. Note: You must be physically present in the classroom.

Great little video about astronomer Jocelyn Bell. She talks of the sexism she endured, speaking with conviction but also with humor. NOT required for our class, but watch it! Only about 15 minutes.

New reading assignments added to list.

We likely will have at least one reading from The Economist. You can access it online via the UCD Library. Go here for the details. (To start the process at economist.com, click on the triple-bar menu symbol, not "Subscribe.")

BTW, here are some other publications you can subscribe to (or otherwise access) using the UCD Library:

• NYT
• San Francisco Chronicle

Most evenings I watch the 10 pm news on Channel 2 in the Bay Area. As with most news shows, they have two co-anchors.

Last night, though, I was taken aback by the fact that both anchors were men. I'd never seen that before, and oddly, it made me feel a little uncomfortable. Say what one will about affirmative action and diversity programs, my reaction tells me that I take a diverse news crew as the norm, which I consider a good thing.

Yesterday, an important question arose in class: Should the definition of disparate impact be limited t) intentional action? (Note the word should, rather than is.) I made an assignment of this; be prepared to be called upon on Monday. There are no "correct" answers, just soliciting your views.

I'll push the second reading for tomorrow (Bao et al) back to Monday, and require only the first part.

I've updated the reading list.

I've moved the New Yorker article to Wednesday's reading.

Link fixed in the 0910 post.

Again, the answers to Exam 1 are on our our Web site.

Reminder: Exam 2 will be held on the last day of class, March 11, during the lecture time slot, 12:10-1. It will cover the articles discussed in class after Exam 1. It's possible that a question in Exam 2 may make reference to something covered before Exam 1, but not in any detailed way.

I just mailed out your Exam 1 grades. You can find the answers in the OldExams directory on our Web page. Overall, the results were good, though with one or two very low scores that very much surprised me.

Around the seventh week or so, I will send you my records in terms of your responses to the Random Questions and your class participation.

Very sorry I haven't yet updated the reading schedule. For tomorrow, please just read the "...Hardly Ever Is" article. I have a question to put to all of you (separate from the article).

Two (unrelated) items:

• As Stefan said, remember that tomorrow's exam will be given during the discussion section, not the lecture.
• If you wrote your name card with an ordinary ballpoint point, please write over the letters with a wide-tipped pen, e.g. felt-tipped, so that I can see it well at a distance.

As Stefan said, make SURE you have actually gone through the OMSI process, so that you will be able to take and submit Monday's exam.

The OMSI docs say the OMSI window must fill your screen at all times. For our course, though, this would be infeasible. So, you may access our course materials during exams, but be sure not to have e-mail or other forms of communication on your screen.

Lest anyone get the wrong impression, although I do feel the "canceling" of Ronald Fisher was unfair, I definitely do not subscribe to his views on encouraging (in any form) the reproducing among so-called "good" families.

Happy Chinese New Year to all! You don't have to be Chinese to enjoy this fun holiday.

As noted, I have shifted the reading scheduled for Wednesday and next Monday back by one lecture each. For Wednesday, read the new assignment, on R.A. Fisher. Rather long, but this is the only item for Wednesday; please read it in full.

As some our course topics are rather sensitive regarding race, ethnicity and the like, I think it would be good to summarize my own background, at least the generally publicly known aspects.

I was born and raised in southern California. Until I was 9, we lived in the City Terrace neighborhood of East LA, which was about equal parts Latino, Asian and Jewish. I was in the latter group, and my parents spoke Yiddish to each other. (Alas, my brothers and I only know a smattering of the language.) My dad was an immigrant from Lithuania; Mom was a US native, but her parents were immigrants.

When I was 9, we moved to the city of West Covina in the San Gabriel Valley, a mainly white, lower-middle class area. I was the first in my family to go to college; I commuted from home to Cal Poly, Pomona, 9 miles away. For grad school, I again stayed in the area, attending UCLA.

Money was very tight in my family. Without going into details, I'll just say for example that I had to borrow money to take the SAT. I had my first-ever airplane ride at age 26. Guess where it went to! Yep, Sacramento, the occasion being my job interview at UCD.

As I mentioned in class today, my wife is a Chinese immigrant, from Shanghai and Hong Kong. Our "official language" at home is Cantonese. At extended family gatherings, the languages Cantonese, Mandarin and Shanghainese are all used with approximately equal frequency.

Among other things, it's worth pointing out that I've lived in immigrant households my entire life. I know the feeling of eating different foods, celebrating different holidays etc. from the mainstream, a general feeling of being The Other.

I've been active in promoting the well-being of people of color and women all my life, starting with activity in high school in the campaign for a Black woman for Congress. (She lost.)

More details are in my bio .

I am adding to our reading list this article on Fisher and eugenics. Please read it to prepare for Wednesday's class. (I'll push the previously-scheduled material back.)

The other day, I noted that in the readings schedule, I moved to a weekly listing rather than daily. This gives us more flexibility, but does mean you'll need to read a week's worth of aricles at time.

Tomorrow I will cover the "10-page ML" book, and possibly start the Strubell article.

(Ghosted by Stefan.)

---

As mentioned in the syllabus, exams this quarter will be conducted on OMSI. We won't have an opportunity to do a dry run of OMSI during a class period this quarter, so there will be a virtual dry run and you all will be responsible for testing it on your own. Failure to submit from OMSI during the actual exam may result in an F grade for the exam.

To that end, Stefan will be setting up a submission server on Thursday morning and will leave it up until the evening so you can practice submitting. Details for the submission server will be sent out on Thursday. Stefan will be reachable by email to answer any questions you might have while testing the software. A couple of useful tips:

• Make sure the Python version you are using is Python 3. (You do not need R.)
• First run your own private test on your machine (localhost), playing both the role of the instructor and student, checking in the end that your submission did indeed reach the server in the proper directory.
• Make sure you are connected to the library VPN when trying to connect to Stefan's server.

I think you've all received the Chancellor's message, re in-person instruction starting Jan. 31. All lectures and discussion sections will be held in-person. In particular, our Feb. 7 exam and the student presentations, both of which are held in discussion sections, will be in-person.

Very sorry for my delay in updating the reading list. I've done so now.

It's more realistic to have a weekly reading assignment, so I've written it that way. But for tomorrow, just prepare for the climate model and Volkswagen articles.

BTW, if you have a "foreign" name, meaning almost anything other than John or Mary :-) please send me e-mail explaining how to pronounce your name. In the case of those with Chinese given names, giving me the characters would be helpful. Not a big thing, but I would hate to keep mispronouncing people's names.

My grading in this class is very liberal. BUT you have to do the work: reading, two exams, presentation and class participation. If you are missing one of these, you will not get a passing grade in the course.

From Stefan:

From sabroecker@ucdavis.edu Wed Jan 19 13:35:00 2022
From: Stefan Broecker 
Date: Wed, 19 Jan 2022 13:34:47 -0800
Subject: 188 Presentations
To: Norm Matloff 

Hi Dr. Matloff,

Could you send out a reminder to students in 188 that there are still
students who have unapproved topics in the spreadsheet? I emailed everyone
on the signup sheet whose topic wasn't accepted, and only a handful of
students have gotten back to me. Everyone on the spreadsheet in either red
or yellow is still unapproved and should email me.

You can also let the class know that there are now two open spots due to
students dropping the class. Some of the students who missed the first
round of slots might be able to get one now.

Best,
Stefan

The link.

I've revised our reading schedule, pushing the census data article back to Friday, and adding a new reading to serve as background.

I have slightly updated our reading schedule. Note the addition of a new reading from the WSJ tomorrow. I will also add a new reading on privacy.

Stefan has set up his presentation grading rubric. Also, his slides from today are here.

It mentions in our reading list that some items may be viewable only through the UCD Library portal. Our "Obsolete Science" reading is one such item, accessible here.

There appear to be a couple of students who have not signed up for presentation time slots. Please note that all components of the class are integral parts of the course. Failing to make a presentation is like not showing up for the final exam in an ordinary course. You must make a presentation in order to get a passing grade in the course.

Please remember to have an example of real-world use of p-values for class tomorrow. The first few Random Questions that I ask will consist of asking the student to very briefly (say 30-60 seconds) describe the setting and conclusion, as well as the source (e.g. if academic journal, name of the journal, first author, date of the journal issue).

I've added a new item to our reading list (Jonathan Turley, Jan. 12), and set the schedule for that day.

Great example of misleading graphics on Axios. Contrary to the claim that certain states are hotbeds of insurrectionism, the more populous states have more cases, by and large. Normalizing by state population may find some statewide differences, but much less so than the graph implies.

The key to the problems with hypothesis testing is that it asks the wrong question. The null hypothesis, H₀ is basically always false. The departure from H₀ may be extremely tiny, but nonzero.

To simplify the discussion, I like to use a coin example. Say we have a coin that we will toss to see who gets the kickoff in the Super Bowl, something like that. We want the coin to be fair, so we have H₀: p = 0.5, where p is the probability of heads for the coin. We toss the coin n times to test whether H₀ is true,

Say we take n = 100 and it turns out that we get 68 heads. If H₀ were true, what would be the probability of getting so many heads? This is the p-value. The philosophy is: If the p-value is really small, our data would have been very unlikely (again, if H₀ were true). Rather than believe we observed a small-probability event, we choose to disbelieve H₀. We would announce, "This coin is significantly unfair."

But the word "significantly" doesn't mean what it does in ordinary English. It doesn't mean that the coin is unsuitable for use in the Super Bowl.

Say for instance the true heads probability is 0.502. Good enough for the Super Bowl, most would say. But if n is large enough, our sample proportion of heads will be close to 0.502, and that proportion would be unlikely of p = 0.5. So for large n, we'd reject H₀ and say the coin is "significantly" unfair, when if fact that coin is fine.

Again, the problem is that H₀ is misspecified, postulating a state that we know a priori is false. And if for small n we reject H₀, declaring p to have no significant difference from 0.5, we're making the same mistake.

Sorry, I missed the point brought up in the Chat box, which was, "Why not use the term 'absence of evidence'?", in the context of accepting the null hypothesis. I'll answer in the next blog post, later this afternoon.

Please re-read the blog post of Tuesday, January 4, 10:15 IMMEDIATELY. If you do not specify an article that you will present, YOU WILL LOSE YOUR TIME SLOT.

Cloud recordings of lectures now up to date, sorry for the delay.

Please read the syllabus IN DETAIL, AS SOON AS POSSIBLE, i.e. TODAY.

Please note carefully that when you sign up for a presentation time slot, you must state (a) the URL of the Web entity that you will be presenting and (b) the URL of the item in our reading list that your presentation will relate to.

If you have already signed up, please add this information SOON in order to RETAIN the slot you signed up for!

I've added my office hours to the syllabus, MW 10-11, 3053 Kemper. They will begin next week, but if you need to discuss something with me before then, feel free to send me e-mail. We could also meet by Zoom if you prefer.

I've updated our syllabus to show: (a) Our first exam will be on February 7 (in discussion section). It already mentioned that the second one will be on March 11, in lecture. (b) I've posted Stefan's signup sheet for presentation time slots. HURRY! to get the one you want.

Your reading schedule is here.