Wednesday, July 2, 2025
Context: In my role as division director of Information and Intelligent Systems (IIS) at NSF, I’m sending out a short message to the IIS mailing list on the Second Tuesday Every Month (STEM). This is the installment for July 2025, sent a little early because I’m about to lose access to the list.
Hi all,
When I first arrived at the NSF, I was overwhelmed by all the acronyms. Of course, acronyms are common in all of areas of intense study, and my life in higher education certainly exposed me to many. (POMDPs are one of my favorite RL topics!) But the rapid culture change and the fact that the NSF is itself the collision point between the technical jargon of academia and the bureaucratic lingo of government meant that I was suddenly expected to know what people meant when they asked for an update on the NITRD AI R&D IWG SP. (That’s the Networking and Information Technology Research and Development Artificial Intelligence Research and Development Interagency Working Group Strategic Plan, something I worked on when I arrived and helped kick off on my way out.)
I started keeping track of acronyms that people used with me without explanation, mostly as a way of not having to ask for explanations twice, but also as a kind of vindication for my belief that there are a lot to them to know. Indeed, my list had 90 entries in my very first week. Note that I am using the word “acronym” in an inclusive way, as some of these entries might more properly be called initialisms (like IIS for the Information and Intelligent Systems division, where each letter is pronounced), or abbreviations (like ENG for the Engineering directorate, pronounced “ehnj”), or nested acronyms (like CRII for the CISE Research Initiation Initiative program, which includes CISE for the Computer and Information Science and Engineering directorate), or backronyms (like CIVIC for the Civic Innovation Challenge program, where the name likely came first and then the acronym was created to fit it), or classic pronounceable acronyms (like OLPA for the Office of Legislative and Public Affairs at NSF).
In the three years of collecting acronyms, my list reached 1,091 entries. If you are curious, you can find it on my website here. I categorized each acronym by the type of entity it refers to:
361 organizations (like AF for the Algorithmic Foundations cluster that funds theoretical computer science work in CISE, non-ironically pronounced “A F”)
261 science terms (like VR for virtual reality, often encountered in the context of grant proposals)
147 examples of administrative terminology (like RSV for Reverse Site Visit, which were neither reverse, nor a site visit, during my time at NSF)
130 programs (like SCH for Smart Health and Biomedical Research in the Era of Artificial Intelligence and Advanced Data Science, originally Smart and Connected Health, pronounced “smart health”)
67 roles (like CRSSP for Chief of Research Security Strategy and Policy, the NSF person overseeing research security topics, pronounced “crisp”)
57 documents (like GAIRA for the Global AI Research Agenda, pronounced “GAIR-uh”)
23 software-or-systems-related terms (like WAVES for the White House Worker and Visitor Entry System)
18 places (like NOVA for Northern Virginia, where NSF has been located since moving from DC in 1993)
17 laws (like GPRA for Government Performance and Results Act, a law with the goal of speeding up government processes, pronounced “GIP-rah”)
7 meetings (like FCRC for the Federated Computing Research Conference)
2 people (like MRM for Margaret Martonosi, the CISE director who hired me), and
1 course (AP-CSA for Advanced Placement Computer Science A)
During my time at NSF, I even got to create a few acronyms of my own, like ACED for the Accelerating Computing-Enabled Scientific Discovery program, pronounced “aced” and AISC for the NSF-wide AI Steering Committee that I co-chaired, pronounced to rhyme with “basic”.
Overall, my acronym list is a kind of time capsule of my tour in the Federal government, nodding to some of the high points (like CMG for CISE Management Group, the folks I met with each week to help oversee the activities of the CISE directorate) as well as some of the more emotionally fraught topics (like DRP for Deferred Resignation Program, the plan that was recently offered to NSF employees that resulted in many talented people leaving the agency with their roughly millennium of combined institutional knowledge). The list is a condensed form of something much larger and more complex, distilled down for easier reference. Kind of like an acronym…
I’ve gotten feedback that some people on the list might be interested in continuing to hear from me even after I transition out of NSF and into my new role. If you’d like to sign up to join this NOT-NSF-APPROVED mailing list, click here. In terms of NSF Approved mailing lists, I highly recommend signing up for Ellen Zegura’s CISE Newsletters. She’s CISE’s acting directorate head and has been doing a great job of providing “news you can use” during these tumultuous times. Past newsletters and a signup link can be found here.
Last STEM, I proposed a puzzle that revolved around figuring out how to reconstruct this week’s Billboard Top 10 list from last week's top ten and the current week's markings saying, for each rank position, whether the song at that rank went up, down or stayed the same. In case the challenge was hard to follow, here’s a concrete example:
List of Top 10 Songs for Week of June 21, 2025:
Manchild [Sabrina Carpenter]
Ordinary [Alex Warren]
What I Want [Morgan Wallen Featuring Tate McRae]
Just In Case [Morgan Wallen]
Luther [Kendrick Lamar & SZA]
I'm The Problem [Morgan Wallen]
A Bar Song (Tipsy) [Shaboozey]
Die With A Smile [Lady Gaga & Bruno Mars]
Lose Control [Teddy Swims]
Beautiful Things [Benson Boone]
Markings of Top 10 Songs for Week of June 28, 2025:
up
down
same
same
same
same
same
same
same
same
In this example, we can logically reconstruct this week’s Top 10, if we know it was the same songs. In particular, starting from song 2, we know that whatever it is, it must have gone down from last week. But to go down to position 2, the song must have been in position 1 last week. So song 2 is Manchild [Sabrina Carpenter]. We know that songs 3 through 10 stayed the same. So, using the knowledge that the songs are a permutation of last week’s list, song 1 is the only one unaccounted for: Ordinary [Alex Warren]. Thus, we've uniquely reconstructed the Top 10.
Here's an example from almost exactly a year ago where we can't reconstruct the Top 10.
List of Top 10 Songs for Week of June 29, 2024:
Please Please Please [Sabrina Carpenter]
I Had Some Help [Post Malone Featuring Morgan Wallen]
A Bar Song (Tipsy) [Shaboozey]
Espresso [Sabrina Carpenter]
Million Dollar Baby [Tommy Richman]
Not Like Us [Kendrick Lamar]
Too Sweet [Hozier]
Beautiful Things [Benson Boone]
Lose Control [Teddy Swims]
Birds Of A Feather [Billie Eilish]
(It's kind of amazing that there are three songs in common a year later! And Morgan Wallen, Kendrick Lamar, and Sabrina Carpenter are back with different songs.)
Markings of Top 10 Songs for Week of July 6, 2024:
up
up
up
same
down
down
same
same
same
same
Here, there are multiple ways the ten songs might be assigned to these positions. One is:
A Bar Song (Tipsy) [Shaboozey]
Million Dollar Baby [Tommy Richman]
Not Like Us [Kendrick Lamar]
Espresso [Sabrina Carpenter]
Please Please Please [Sabrina Carpenter]
I Had Some Help [Post Malone Featuring Morgan Wallen]
Too Sweet [Hozier]
Beautiful Things [Benson Boone]
Lose Control [Teddy Swims]
Birds Of A Feather [Billie Eilish]
But another solution would be to swap the songs in positions 5 and 6. (If I'm not mistaken, there is a total of 13 different solutions.)
So sometimes the exact order can be reconstructed and sometimes it can’t and the challenge was to count the number of permutations that could be uniquely reconstructed.
Alas, no one on the mailing list of about 12,000 people sent me an attempted solution. So maybe it’s not solvable? I guess we’ll never know. (Just kidding. If you join my mailing list, I’ll send you my solution. But maybe these worked examples will help get you unstuck?)
My fondest thanks to the entire NSF organization, the stellar program officers, my fellow division directors and deputy division directors, the remarkable directorate heads and deputies, the dedicated operations staff, the phenomenal communications team, the troopers in the office of the director, and everyone else. It was deeply meaningful to me to get to be a part of the team in our efforts to serve the broader scientific enterprise.
And thanks to all of you… every one of these messages spawned some number of responses and it was treat to hear what folks in the IIS community are thinking about. Some of you sent me some excellent suggestions to help improve NSF, many of which I didn’t get to implement due to lack of … well, multiple things. My apologies. But keep up the amazing work.
It’s been fun, and even when it wasn’t fun, it sure was interesting.
I’m so glad we had this time together!
Michael