Conal Elliott

I am seeking research collaborators and mentees on the topic of correct and efficient hardware and software engineering with *simple* and precise specifications. github.com/conal/Collabor…

@satnam6502 @HaskelInterlude Your work has inspired me, too! I'm always eager to collaborate on bringing elegant, principled math/logic to efficient and useful implementations.

An inspiring chat with @conal on the @HaskelInterlude with a refreshing perspective on how to do hardware design from a disciplined mathematical perspective, coded up in Agda. Also, his comment that "... I knew that matrices are actually not the right way to think about linear algebra" might stop a few machine learning folks in their tracks. If anyone else is doing hardware in Agda then I'd love to hear from you. haskell.foundation/podcast/62/

I feel a lot of pressure to be a graphs and charts computer scientist, a quantitative computer scientist. I hate it, and I've spent most of my career doing exactly the opposite of what Hennessy and Patterson have espoused. I want to be qualitative computer scientist. I am very happy with the subjective rather than so called objective. I believe there is value in an idea or concept that does not need to be subjected to the Spanish Inquisition of a PLDI paper review. I believe there is truth in beauty, and it does not need to be measured by a Ruler.

@jonathoda @mechadense @disconcision @dm_0ney @neurocy Yes, and as fluid and powerful a theory of change as Newton's & Leibniz's.

@mechadense @conal @disconcision @dm_0ney @neurocy Perhaps we can all agree that we need a better theory of change for both code and data. Transformations, lenses, functors, derivatives, edit calculii, diffs are all constructions of change

📝Future Of Software. A brain-dump. Decently happy with the result. Might be interesting for my CompSci and future-of-coding followers. Specifically: @conal @jonathoda @disconcision @dm_0ney @neurocy with their work more or less directly mentioned. Point me to remnant typos & co.

through a shocking turn of events I am moving back to sf (for biz money reasons). if you know of a cat-friendly 2b/2b in a safe, walkable hood, let a sis know 🙏🏽

My daughter Friday's Kickstarter campaign is down to the last two days: kickstarter.com/projects/frida… .

@mechadense @sabine_s_ @jonathoda Alternatively, specify and correctly implement what you can in DD-Agda, and generate code in a non-correctness-oriented language for interop with non-correct libraries.

@mechadense @sabine_s_ @jonathoda Graphics is a natural field for specifying precisely and implementing correctly (for real, i.e., machine-checked proofs). Of course, HW & SW devs will keep building mountains of incorrectness for a while, but I'm encouraging those who care to start a new ("Galilean") foundation.

dependent types are the Bermuda Triangle of software engineering

@jonathoda @mechadense @sabine_s_ Without addressing correctness, dependent types are yet another fancy tool for incorrect computation.

@jonathoda @mechadense @sabine_s_ Machine-checked correctness, i.e., faithfulness of an efficient and/or sophisticated implementation to a precise and simple (thus non-operational) specification.

@mechadense @sabine_s_ @jonathoda A philosophical analysis of indifference to truth: en.wikipedia.org/wiki/On_Bullsh… .

@mechadense @sabine_s_ @jonathoda Lack of sophisticated types/distinctions in mainstream programming languages is a symptom of a more fundamental failing: lack of commitment to and discipline for truth/correctness in mainstream computing education and practice.

@mechadense @sabine_s_ @jonathoda Agda is much more practical than you realize, and is the *most* practical language I know for the ("Galilean") computational revolution from guessing to knowing. (You might have some limiting assumptions about how to use Agda.)

@sabine_s_ @jonathoda @conal I wish there were practical media focused languages like elm supporting deptypes. Using deptypes in practical environment rather than borderline research language or as an eventual frankenstein tack-on onto more mainstream lang (e.g. haskell). And please no church numerals 🤦.

@mechadense @sabine_s_ @jonathoda Dependent types enable us to take responsibility for thinking and computing correctly. However, "As scarce as truth is, the supply has always been in excess of the demand." - Josh Billings

@mechadense @sabine_s_ @jonathoda I practiced Denotational Design for decades without dependent types and adopted them because my mentees needed formal support for specifications and proofs. Dependent types matter because logic matters because truth matters. #readme" target="_blank" rel="nofollow noopener">github.com/conal/talk-202…

@p1xelHer0 Next thing: *simply correct computing*. github.com/conal/Collabor…

I gotta admit I am a bit tired of people saying: - Declarative - Easy to reason about - Simple And all that jazz. I am already FP brainwashed I need the next thing. Ascend!

Einstein slept nearly 10 hours a day. He valued the significance of quality sleep for maintaining his well-being. His sleep routine, though, was out of the ordinary. He regularly indulged in brief daytime naps, occasionally more than once a day. These quick naps, lasting just a few minutes, were thought by Einstein to revitalize his mental energy and creative thinking. He felt that these short periods of rest gave him a renewed viewpoint to solve intricate scientific challenges.

If you've been wired to optimize for achievement, actually leaning into the "real work" you are meant to be doing can feel really bad. This is because you have to mourn the loss of identity you had tied to achieving other people's goals. Not to mention the time it takes to lean into what you are truly meant to do. I think one of the biggest scams of modern work culture is the idea of "Ikigai" - it's this made up western interpretation of a Japanese word and includes a diagram convincing people that "what you can be paid for" is a reasonable thing to expect from work I think if you want to find something only you can do, you must completely release attachment from any sort of outcome, impact, money, respect, etc... In clinging to these expectations, we often take shortcuts to solve for those rather than deepening our connection to the work we are meant to do. The world convinces far too many (including me in my twenties) that we CAN have it all. But the truth is that you can't be anything you want to be you can only be what you truly are. We think doing your own thing is this grand celebration of autonomy and self-determination, not to mention an inevitable arrival at achievement But from what I've seen that only comes sometimes after someone releases to the tragic reality that our interests, curiosities, desires, and proclivities don't fit into an achievement-shaped path. Some people are wired for this of course but its quite hard to tell how many people are deluding themselves and how many are truly meant to be doing that thing. The start of a path that feels good is often filled with sadness and mourning of a past identity, one that got you where you are but won't help you be the person you know you can be.

Here's where we are leading up to SPLASH 2023! We have seen an extended global reach for mentees, primarily in South America. We have also seen growth in the number of mentees, but the number of mentors have been stable. We are badly in need of new mentors. Please sign up!

The proceedings of ICFP 2023 - PACM PL, Vol. 7, Issue ICFP has now been published in the ACM Digital Library: dl.acm.org/toc/pacmpl/202…