Advertisement · 728 × 90

Posts by Mike Samuel 🟣

congrats!

38 minutes ago 0 0 0 0

No

3 hours ago 0 0 0 0
[Screenshot from gmail]

Jason Martin
to icon-language

After kidney cancer, double bypass and eleven years on this project, I achieved most of my goals.

https://github.com/agrellum/gorexxr

[Screenshot from gmail] Jason Martin to icon-language After kidney cancer, double bypass and eleven years on this project, I achieved most of my goals. https://github.com/agrellum/gorexxr

I haven't checked in on Icon land lately. I wonder what's going on.

...

well, shit.
also, good on you.

3 hours ago 1 0 0 0

Sometimes the latter can be both intensely social and also very brusque because they recognize that their time is a scarce resource for the org as a whole.

1 day ago 1 1 1 0

Yeah. I know two kind of senior staff, both of them very social in quite different ways.

The uplifters you describe.

And infrastructure maintainers who build something that many others build upon, who actively supports their building by engaging with a huge number of people outside their team/org.

1 day ago 3 1 2 0

Yeah. There was a very wide gap between senior and staff where I crossed over.

The criteria didn't explicitly list mentoring (though I had been mentor to a guy who gave half of the "so you want to be an eng mentor" trainings) but it required project work that would be tough to pull off purely solo.

1 day ago 1 1 1 0

There's a dynamic around mentoring effort too.

One can promote to senior just by time in grade or technical savvy.

A senior has to intentionally tamp down "I could work with a junior to get this done, but it'll be faster to do it myself" until they're skilled enough at transfer to default to it.

1 day ago 3 1 1 0
Advertisement

a Yiddisher Khedzman

2 days ago 1 0 0 0

@anil.recoil.org

3 days ago 1 0 0 0

Yeah, baking and candy making are just different.
I suspect every aspiring, amateur confectioner goes through exactly that frustration.
For me, I didn't get it until I sat down and made gâteau Basque by rote, no vibes, until it came out right.

4 days ago 0 0 0 0
His enthusiasm for Arthur is apparent in the work. The book was left unfinished at his death, and ends with the death of chivalry in Arthur's purest knight, Lancelot of the Lake.[2]: Chase Horton, Appendix, p. 296. 

His enthusiasm for Arthur is apparent in the work. The book was left unfinished at his death, and ends with the death of chivalry in Arthur's purest knight, Lancelot of the Lake.[2]: Chase Horton, Appendix, p. 296. 

Apparently it's incomplete which I didn't notice reading it as a child, but he's a great writer and his enthusiasm does come through.

en.wikipedia.org/wiki/The_Act...

4 days ago 0 0 0 0

Nice!
I love Indian saucery since you can get great results just vibing, but baking requires a level of precision.

My 10yo daughter has decided she's vegetarian and is butting heads with my Telugu mother in law. I'm here thinking: the protein is just lentils, what are y'all even arguing about.

4 days ago 2 0 2 0

I very much liked Steinbeck's translation of Le Morte d'Arthur but I don't really know the origins. There's a lot of crossover between Celtic, Germanic, and French folk traditions.

In Welsh mythos, Math is the "Hey everybody, I know how to solve this" guy like Merlin in the Arthurian tales.

4 days ago 0 0 1 0

What all do you cook?

4 days ago 1 0 1 0
Advertisement
4 days ago 4 0 1 0

If all wheeled populations die out early perhaps due to historical accident, and there is no mostly positive quality path from finely tuned legged bodies to poorly tuned wheeled bodies, the algorithm isn't going to magically shepherd a wheeled population into existence.

4 days ago 1 0 0 0

Imagine that you, a god, are using evolutionary algos to try to design the perfect robot for some task.
Perhaps wheels are objectively the best means of locomotion, but the algorithm also tries legs which, at the scale of early populations, do a better job navigating seams in pavement.

4 days ago 1 0 1 0
Hopper [1999] vividly described this viewpoint. “Progress in evolution depends fundamentally on the existence of variation of population. Unfortunately, a key problem in
many Evolutionary Computation (EC) systems is the loss of diversity through premature convergence. This lack of diversity often leads to stagnation, as the system finds
itself trapped in local optima, lacking the genetic diversity needed to escape.” Until

Hopper [1999] vividly described this viewpoint. “Progress in evolution depends fundamentally on the existence of variation of population. Unfortunately, a key problem in many Evolutionary Computation (EC) systems is the loss of diversity through premature convergence. This lack of diversity often leads to stagnation, as the system finds itself trapped in local optima, lacking the genetic diversity needed to escape.” Until

The AI literature itself has plenty of discussion about the tendency of algorithms that mimic evolution to stop effectively progressing.

romisatriawahono.net/lecture/rm/s...

4 days ago 2 0 1 0

Here I am in my second decade of my one man crusade to bring "whence" and "whither" back into common usage, but everyone's just going to talk like AI.

It is sad, it is not happy making. 😭

4 days ago 1 0 0 0

I worry about the effect of that messaging on education.

Will kids not using AI for schoolwork get left behind?

Probably not, but ed policy debates deeply confuse parents, high-production-value messaging only lines up behind products, and learning by the sweat of one's own brow is not a product.

4 days ago 9 0 1 0

Brahmin? Of course I know what that means: rich white people from Boston.

4 days ago 1 0 0 0
Advertisement

Don't become Blodeuwedd-pilled.

5 days ago 1 0 0 1
Video

Trump has threatened to use the U.S. military to carry out acts of genocide against the Iranian people.

Make no mistake: those that voted NO to the War Powers Resolution are actively endorsing that threat.

Congress must exercise our power to end this chaos.

5 days ago 1122 352 21 9

So say we all

5 days ago 1 0 0 0

Functional completeness breaks brains.

I think people get invested in the idea "I worked hard to learn these 10 operators so each of them must be a jewel of distinct splendour" and, no, redundancy in math notations is subtextually important like everywhere else.

bsky.app/profile/mvsa...

5 days ago 0 0 0 0
Welcome! I'm your friendly neighbourhood first code reviewer.

First reviews can come with uncertainty so I thought I'd reach out and
explain the process.

The first review often takes longer than average, and there's a
process for fast-tracking reviews when something is down, or there's
extraordinary time sensitivity.
(See links at the end, or chat me and I can expedite)

First rule of code reviews around here, don't commit until you get a
"+1." That's the explicit signoff, and it just helps avoid confusion
and miffed feelings.

Second, you don't have to do what I say. We tend to run on consensus,
so if you and I just don't agree, we'll loop in someone else. It is
good to respond to every point raised inline, even if it's just
"addressed above."  That way it's clear who is waiting for whom on
what.

Third, you can drive the kind of review this is. Just let me know
where you are in the design -> rollout process. If you're sketching
something out, I can focus on design advice, if you're in some really
thorny code, I can suggest tests for corner cases, if you're working
in a language you haven't worked on, or plan to apply for readability,
I can be particular about how we tend to style things here so the
codebase as a whole is easy to read.

Finally, if you don't want me as a reviewer for any reason, I'm happy
to just throw it back in the queue for the auto-assigner, no questions
asked.

Welcome! I'm your friendly neighbourhood first code reviewer. First reviews can come with uncertainty so I thought I'd reach out and explain the process. The first review often takes longer than average, and there's a process for fast-tracking reviews when something is down, or there's extraordinary time sensitivity. (See links at the end, or chat me and I can expedite) First rule of code reviews around here, don't commit until you get a "+1." That's the explicit signoff, and it just helps avoid confusion and miffed feelings. Second, you don't have to do what I say. We tend to run on consensus, so if you and I just don't agree, we'll loop in someone else. It is good to respond to every point raised inline, even if it's just "addressed above." That way it's clear who is waiting for whom on what. Third, you can drive the kind of review this is. Just let me know where you are in the design -> rollout process. If you're sketching something out, I can focus on design advice, if you're in some really thorny code, I can suggest tests for corner cases, if you're working in a language you haven't worked on, or plan to apply for readability, I can be particular about how we tend to style things here so the codebase as a whole is easy to read. Finally, if you don't want me as a reviewer for any reason, I'm happy to just throw it back in the queue for the auto-assigner, no questions asked.

When managers wonder about the value of CR, I share an email template that's followed me through a few jobs.

Most managers hesitate to publicly discount things like onboarding, documenting designs, or test coverage; but the link between those and code review (done well) isn't always clear.

5 days ago 4 1 1 0

Nice. The `subleq` of real arithmetic.

6 days ago 1 0 0 0

The fact that people think that their dislike of Piker, however justified, then means they shouldn't care about this deliberate misinformation and reliance on bigoted sources, betrays a deep bias.

It is this lack of concern that generally leads to the further spread of disinformation.

6 days ago 220 21 1 0
Advertisement

Yeah. In one case we can point to beliefs causing things but not the other, e.g. money can be traded for goods and services today because people believe money will continue to be tradable for goods and services tomorrow.

6 days ago 2 0 0 0
OWASP AppSec 2010: Beyond the Same-Origin Policy 1/3
OWASP AppSec 2010: Beyond the Same-Origin Policy 1/3 YouTube video by Christiaan008

@holobrine.bsky.social, relevant to OCaps for AI, here's a talk that @jasvir.bsky.social and I gave in 2010:
Virtualization as a meta-tool.

Designing APIs that you can virtualize gives you flexibility in evolving those APIs without breaking compatibility.

www.youtube.com/watch?v=FJHj...

1 week ago 2 0 0 0