Unpacking models of what the $USD1.3B might achieve in Computing Education: We need long-term vision and will

I wrote my Blog@CACM post for September on the massive investments in CS Education announced last week (see post here): $200M/year from the US Department of Education announced by the White House on Monday, then $300M over five years from the Tech industry announced on Tuesday. I have read analyses saying that the money isn’t really promised or isn’t new (see concerns in this post), and others are shunning the initiative because of White House policies (see link here). I took the promises at face value. My post starts congratulating Hadi Partovi and Cameron Wilson of Code.org and Ivanka Trump who were behind these initiatives, then I offered two back-of-the-envelope models of what $1.3B in five years could do:

  • I extrapolated the New York City model (of a significant computing education experience to every child in every school within grade bands) to the whole of the US, which would likely take more than a magnitude more funding.
  • The funding is enough to pay for a CS teacher in every school, but I argued that it wouldn’t really work. We face a shortage of STEM teachers, and those few are the teachers that we can most likely recruit to CS. CS teacher attrition is so high that we couldn’t keep up with the losses, since we have so few mechanisms of pre-service CS teacher preparation.

I received many responses, queries, and criticisms of that blog post (from email, Facebook, and Twitter).  I am explaining and unpacking the CACM blog post here. I am not going to delete or change the CACM blog post. My mentor, Janet Kolodner, told me once not to dwell on any paper, trying to make it a masterpiece before publishing it. Rather, she suggested that we should just keep publishing. Explore lots of ideas in lots of papers, and publish as a way of thinking with a community. It’s okay to publish something you thought was right, and later find that it’s wrong — it documents the explored trails.

What I learned about the effort in NYC

I said that NYC was aiming to provide a quality computing learning experience for every student in every grade in every school, as I learned last October (and blogged about it here). I learned that the goal is now mandating a computing learning experience in every grade band, so not every year. It’s still a markedly different model than one teacher per school, and doesn’t change the costs considerably.

I learned that (as one might expect) that the effort in NYC is in both the NYC Department of Education and in CSNYC. It’s great that there are many people in the NY DoEd working on CS education! I was told on Twitter that some of what I attributed to CSNYC is actually in NY Department of Education. I don’t know what I mis-attributed, but I’m sure that it’s because I get confounded over “CSNYC” representing “the effort to provide CS education across NYC” and “the organization that exists to provide CS education across NYC.” I don’t understand the split between NYC DoEd and the CSNYC organization, and I’m not going to guess here. I am sure that it’s important for the people involved, but it’s not so important for the model and national analysis.

Explaining my Estimates in Contrast to Code.org’s

Code.org has made their model of the one-time cost of expanding access to K-12 computer science (CS) available at this Google doc. According to their model, it’s clear that the $1.3B is enough to make CS education available in every elementary and secondary school. They have more empirical data than anyone else on putting CS in whole districts, and their data suggest that costs are decreasing as they gain more efficiencies of scale.

Hadi challenged several points in my blog post on Facebook. I won’t replicate all of our exchange, and only include three points here:

  • I argue that we will probably have to pay future CS teachers more in the future, at least as teacher stipends. That prediction is based on trends I see in the states I work with and economics. States are facing teacher shortages, especially in STEM. Aman Yadav shared an article (see link here) that students studying to be teachers fell by 40% from the 2010-2011 academic year to the 2014-2015 academic year. If the supply of teachers is growing more slowly than the rate at which we’re trying to grow CS, we will have to provide incentives to make CS more attractive. Lijun Ni’s dissertation explored the barriers for teachers to become CS teachers (e.g., it’s a lot easier and more pleasant to stay a math teacher). Costs are likely to grow as the labor shortage increases.
  • Some of my costs are too high, e.g., I estimated the cost to develop a high school CS teacher as $10K, where NSF’s studies found it was closer to $8.6K. I used a ballpark 50% of high school CS teacher development for the costs of elementary school CS teacher development.  Since it’s clear that there is enough to prepare one CS teacher per school, I think my numbers are close enough.
  • I believe that extrapolating the NYC model across the country would be even more expensive than it is in NYC. Travel costs in NYC are much less than in rural America. While NYC is very diverse, the rest of the United States is just as diverse. I got to see Ann Leftwich at Indiana University on Saturday. She told me that some of the schools she works with resist teaching science at all! It’s really hard to convince them to teach CS. I expect that there is a similar lack of will to teach CS across the US.

Not all of my estimates are research-based. We don’t have research on everything. Changing all US schools happens so rarely that we do not have good models of how it works. I don’t think that the empirical data of what we have done before in CS Ed is necessarily predictive of what comes next, since most of our experience with CS Ed at-scale is in urban and suburban settings. Getting everywhere is harder. I have observed about “Georgia Computes!” — 1/3 of the high schools in GA got someone that Barbara trained in CS, and that’s likely the easiest 1/3. The next 2/3 will be harder and more expensive.

What I Missed Entirely

As Hadi correctly called me on, the biggest cost factor I missed is the development of curriculum. Back in July, I blogged about Larry Cuban’s analysis that suggested that we need to re-think how we are developing and disseminating CS curriculum in the United States (see link here). We have to develop a lot more curriculum in collaboration with schools, districts, and states nationwide. The US will never adopt a single curriculum nationwide for any subject — it’s not how our system was developed, and it’s why Common Core did not reach all 50 states. The US education system is always about tailoring, adapting, and working with local values and politics. Curriculum is always political.

Mike Zamansky just posted a blog post critiquing some of the curriculum he’s seeing in NYC (see post here). I don’t agree with Mike’s post, but I wholeheartedly agree with his posting. We should argue about curriculum, negotiate what’s best for our students, and create curriculum that works for local contexts.  There is going to be a lot of that nationwide as we take steps towards providing computing education to all students. The iteration and revision will be expensive, but it’s a necessary expense for sustainable, longterm computing education.

What should we do with the money

At a talk I gave at Indiana University on Friday, Katie Siek asked me my opinion. What do I want to see the funding be used for?

It would be great if some of that funding could start more pre-service CS teacher preparation programs. I have argued that we should fund chairs of CS Education in top Schools of Education (see post here). Germany uses this model — they create CS Education professors who will be there for a career, producing CS teachers, supporting local communities of CS teachers, and serving as national models. An endowed chair is $1-3M at most universities. That is not very expensive for a longterm impact.

I prefer an NYC-like model of reaching every student to the model of a teacher for every school. The data I’ve seen from our ECEP states suggests that most CS teachers teach only a single computing class, and that class is typically mostly white/Asian and male. One CS  teacher per school doesn’t reach all the female and under-represented minority students. Equity has to be a top priority in our choices for these funds, since CS education is so inequitable.

My greatest wish is for computational literacy to be woven into other disciplines, especially across all of STEM. I devoted my career to computing education because I believe in the vision of Seymour Papert, Cynthia Solomon, Alan Kay, and Andrea diSessa. Computational literacy can improve learning in science, mathematics, art, language, and other disciplines, too.

I don’t argue that computer science is more important than other STEM subjects. Rather, computing makes learning in all the other STEM subjects better.

I want us to teach real computational literacy across subjects, not just in the CS class hidden away, and not just in an annual experience. I recognize that that’s a long-term, expensive vision — probably two orders of magnitude beyond the current initiative. We need more long-term thinking in CS education, like building up the CS teacher development infrastructure and making the case to people nationwide for CS education. We are not going to solve CS for All quickly.

When the K-12 CS Framework effort launched back in 2015, I told the story here about a conversation I had with Mike Lach (see post here). He pointed out that the last time we changed all US schools, it was in response to the Civil Rights movement. That’s when we started celebrating MLK Jr Day and added African-American History month. He asked me to think about how much national will it took to make those changes happen. We don’t have that kind of national will in CS education in this country — yet. We have a lot more groundwork to do before we can reach CS education for all students or all schools, and funding alone is not going to get us there.

October 4, 2017 at 7:00 am 5 comments

Preparing Tomorrow’s Faculty to Address Challenges in Teaching Computer Science

I’ve blogged here when we have opened registration for the New Computing Faculty workshops (e.g., here), but I haven’t really explained why we’re doing them.  We took a lot of grief on Twitter for the workshops in the Spring, and 120 characters just isn’t enough to explain the whole story. We (Leo Porter, Cynthia Lee, Beth Simon, and me) wrote an article that appeared in the May CACM explaining the rationale.  If you don’t have ACM Digital Library access, you can grab the paper from my Guzdial Papers page here in the blog.

The new challenges compound existing teaching-related challenges for the field. We still need to broaden participation in our field, with the lowest percentage of women majors in all of STEM. The economic rewards of a computing career make it even more important to bridge the digital divide. If there are more students than faculty can teach effectively, they may be inclined to lean on a pessimistic belief that success is dependent on “brilliance” and innate ability where only a subset of students can succeed. If CS faculty feel there is little they can do to change students’ outcomes in their individual classrooms, it will be true. Research shows that more CS faculty hold this mistaken and unproductive view of students than faculty in other STEM disciplines.

Source: Preparing Tomorrow’s Faculty to Address Challenges in Teaching Computer Science | May 2017 | Communications of the ACM

October 2, 2017 at 7:00 am Leave a comment

Developer Bootcamps and Computing Education: Tech Done Right Podcast

I was so excited to be invited to do this podcast with Noel Rappin (my first PhD student) and Jeff Casimir who runs the Turing Academy bootcamp. I learned a lot about bootcamps from Jeff, whom I was pleased to learn is a data geek and measures things pretty carefully.  Two of my favorite insights:

  • Female students are more likely to graduate from the bootcamp. They are more likely than male graduates to leave before six months on the job.
  • Students who skip college and go straight to bootcamp (as Peter Thiel encourages students to do) have a harder time graduating and getting a job. That latter part might be ageism, bias against younger job-seekers.

I recommend the podcast — we had a fun discussion.

How do people learn computing? Who learns best from traditional computer science education and who from bootcamps? How can we teach people who are not developers but who need to learn some programming to do their jobs? Jeff Casimir, the founder of Turing academy, and Georgia Tech’s Mark Guzdial, one of the founders of the International Computing Education Research conference, join Noel to answer these questions and also explain why Excel is both the best and the worst thing in the world.

Source: Tech Done Right Episode 20: Developer Bootcamps and Computing Education with Jeff Casimir and Mark Guzdia

September 29, 2017 at 7:00 am 3 comments

White House announces $200 million a year for computer science – Code.org #CSforAll

Looking forward to hearing more details at Code.org’s webinar this afternoon.  Hadi Partovi posted on Facebook that the money will be provided as competitive grants to schools and non-profits through the Department of Education.  Hadi has written a personal blog post about his motivations in supporting this announcement.

The White House memorandum on the announcement is here. I don’t understand all the details here, and the details of the funding are important.  If it’s not new funding, then it puts CS in competition with other fields, e.g., if the money is set aside for CS when it was originally allocated for all of STEM.  The White House memorandum says, “Establish promotion of high-quality STEM education, with a particular focus on Computer Science, as a Department of Education priority.” If it’s a preference (e.g., a school gets money if and only if they’re teaching CS), it may hurt schools that can’t afford to teach CS yet because they’re stretched thin teaching literacy and mathematics.

Here’s the webinar information: (9/26) at 11am PT, 2pm ET
By web: https://code.zoom.us/j/783490509
By phone: US: +1 646 558 8656 or +1 669 900 6833
Webinar ID: 783 490 509

Today, the White House announced a $200 million per year commitment to computer science education in America’s schools. Unlike similar proposals in previous years, today’s action delivers funding to schools, immediately. Besides expanding access to computer science in schools that previously didn’t teach it, the funds promise to increase participation by women and underrepresented minorities.This funding will jumpstart efforts to ensure every student in every school has the opportunity to learn computer science as part of a well-rounded education. For advocates of increased access and diversity in CS, this is the culmination of years of momentum that began in classrooms, spread to entire school districts, and won the support of business leaders and elected officials globally.

Source: $200 million a year for computer science – Code.org – Medium

September 26, 2017 at 7:00 am Leave a comment

The Negative Consequences of Brown v Board of Education: Integrating Computing Education

The second season of Revisionist History has just finished.  This season didn’t have the same multiple episodes with tight ties to the issues of education as last season (as I described in this blog post), there was one standout episode that does relate to our issues: Miss Buchanan’s Period of Adjustment.  The podcast deals with the negative consequences of the Brown v Board of Education Supreme Court case that declared that separate was not equal and forced schools to integrate.  The well-documented consequence of the integration was the closing of the schools for African-Americans and the firing of Black school teachers.  Gladwell first considers what the Brown family (named in the case) and the other families in the case actually wanted, and about the longterm impact that even today, there are disproportionately few African-American teachers in the US are African-American — and that leads to impacts on students.

When I studied Brown v Board of Education when I was a graduate student at the University of Michigan, we were taught a negative consequence that Gladwell barely touches on.  Gladwell mentions that there were few jobs for an educated Black person at the time of Brown v Board.  The Supreme Court’s decision, and the consequent firing of Black teachers, was an enormous blow to the African-American middle class in the United States.  Employment was lost at a large scale, and longterm impacts on wealth and prosperity can be measured today.

The connection to computer science education is part of the question of how do we reach everyone and help everyone to succeed.  Today’s computing education is de facto segregated — not in the sense of colored vs white classes, but in terms of only certain demographics are in CS classes and other demographics are not.

  • In many of the high schools we work with, even if white and Asian students are in the school population minority, the computer science classes are mostly white and Asian.
  • English CS classes are almost entirely male, maybe even more than in the US (described here).
  • US undergraduate CS classes don’t seem to be retaining women (blog post here).
  • Code.org classes have are almost half poor students (blog post here), and have excellent diversity (see their Medium post here). What are the rich students taking?  The diversity that Code.org is seeing is not reflected in undergraduate CS (see Generation CS report) which has little diversity and has mostly prosperous students. That’s important because undergraduate CS is the path that most students will take to the IT industry, which is mostly white/Asian and male.

How do we improve diversity in computing education?  Can we avoid a heavy-handed and expensive mandate like requiring CS for everyone? I side effect of requiring everyone to take CS might be that we get all the same kind of CS.  Can we provide equal access to everyone without the negative consequences that Gladwell describes from Brown v Board of Education?

Brown v Board of Education might be the most well-known Supreme Court decision, a major victory in the fight for civil rights. But in Topeka, the city where the case began, the ruling has left a bittersweet legacy. RH hears from the Browns, the family behind the story.

Source: Revisionist History Podcast

September 25, 2017 at 7:00 am 1 comment

The Father Of Mobile Computing Is Not Impressed: The Weight of Redefining the Normal

I have been fortunate to have heard Alan Kay talk on the themes in this interview many times, but either he’s getting better at it or I’m learning enough to understand him better, because this was one of my favorites. (Thanks to Ben Shapiro for sending it to me.)  He ties together Steve Jobs, Neal Postman, and Maria Montessori to explain what we should be doing with education and technology, and critiques the existing technology as so much less than what we ought to be doing.  In the quote below, he critiques Tim Berners-Lee for giving us a World Wide Web which was less than what we already knew how to do.  The last paragraph quoted below is poignant: It’s so hard to fix the technology once it’s established because of “the weight of this redefining of the normal.”

What I understood this time, which I hadn’t heard before, was the trade-off between making technology easier and making people better.  I’ve heard Alan talk about using technology to improve people, to help them learn, to challenge their thinking.  But Alan led the team that invented the desktop user interface — he made computing easier.  Can we have both?  What’s the balance that we need? That’s where Neal Postman and Bertrand Russel come in, as gifted writers who drew us in and then changed our minds. That’s why we need adults who know things to create a culture where children learn 21st century thinking and not oral culture (that’s the Maria Montessori part), and why the goal should be about doing what’s hard — not doing what’s universal, not doing what pre-literate societies were doing.  Alan critiques the iPhone as not much better than the television for learning, when the technology in the iPhone could have made it so much more.

He tosses out another great line near the end of the interview, “How stupid is it, versus how accepted is it?”  How do we get unstuck?  The iPhone was amazing, but how do we roll back the last ten years to say, “Why didn’t we demand better? How do we shuck off the ‘the weight of this redefining of the normal’ in order to move to technology that helps us learn and grow?”

And so, his conception of the World Wide Web was infinitely tinier and weaker and terrible. His thing was simple enough with other unsophisticated people to wind up becoming a de facto standard, which we’re still suffering from. You know, [HTML is] terrible and most people can’t see it.

FC: It was standardized so long ago.

AK: Well, it’s not really standardized because they’re up to HTML 5, and if you’ve done a good thing, you don’t keep on revving it and adding more epicycles onto a bad idea. We call this reinventing the flat tire. In the old days, you would chastise people for reinventing the wheel. Now we beg, “Oh, please, please reinvent the wheel.”At least give us what Engelbart did, for Christ’s sake.

But that’s the world we’re in. We’re in that world, and the more stuff like that world that is in that world, the more the world wants to be that way, because that is the weight of this redefining of the normal.

Source: The Father Of Mobile Computing Is Not Impressed

September 22, 2017 at 7:00 am 3 comments

The challenge of retaining women in computing: The 2016 Taulbee Survey: Supplementary Report on Course-level Enrollment

The Computing Research Association (CRA) has just released a supplement to their 2016 Taulbee Survey report.  They now are collecting individual course data, which gives them more fine-grained numbers about who is entering the major, who is retained until mid-level, and who makes it to the upper-level.  Previously, they mostly just had enrollment and graduation data.  These new data give them new insights.  For example, we are getting more women and URM in computing, but we are not retaining them all.

Except in the introductory course for non-majors, the median percentage of women in courses at each level was either fairly constant or increasing [from previous years]. The most notable increase was in the mid-level course, where the median percentage of women went from 17.4 in 2015 to 20.0 in 2016. The median percentage of women in the upper-level course also increased, from 14.1 to 15.9 percent. We see a slight drop-off from the median percentage of women in the introductory course for majors in 2015 (21.0 percent) to the median percentage of women in the mid-level course in 2016 (20.0 percent), and a somewhat larger drop-off between the median percentage of women in the mid-level course in 2015 (17.4 percent) and the median percentage of women in the upper-level course in 2016 (15.9 percent).  Because the median percentage at each level is for a single representative course, not for all students at that level, some of the differences between levels may be attributable to the specific courses on which the institutions chose to report. Overall, however, this trend of decreasing representation of women at higher course levels is congruent with other data.

Source: The 2016 Taulbee Survey: Supplementary Report on Course-level Enrollment – CRA

September 18, 2017 at 7:00 am 4 comments

Older Posts Newer Posts

Recent Posts

October 2017
« Sep    


Blog Stats

  • 1,438,837 hits

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 5,154 other followers

CS Teaching Tips