Andrew’s talk from SSW last week.
49 Likes
Expected a technical talk, got blasted with inspiration and internet nostalgia instead. Can’t complain!
8 Likes
A very welcome one. Technical talks for sure aren’t the bottleneck in the software-that-works pipeline : ).
1 Like
A pickup artist saying “You should try to make everyone want to be around you because you genuinely care about them”…made me remember this gold nugget:
7 Likes
That infohazard video – wow.
3 Likes
Very nice talk! it seems to me that andrew has ambitions to improve not only software, but also the ones who write it.
1 Like
Love the optimism but can’t help it but surrender day to day to the capitalist machine to maintain my already attained confort status.
I really wish there were other ways to sustain my access to electricity, internet and a computer that don’t involve money, but I don’t see a legal one.
I’ll have to wait for someone to invent OnlyZig 
, meanwhile I don’t have an option besides continue to learn and publish small projects while I feed big_corpo_number_03.
A refreshingly kind, humanist talk. Could have been presented at a Humans Should Work conference too.
The duck animation story is real.
20 years ago I was in a lamb university in a bloody 3rd world country. All the professors suck but they pretend to be good.
So I was the only one used LaTeX and everyone else used Microsoft Word 2003…
Apparently students’ thesis are all gotten into troubles simply because the professors are literally illiterate… When I submitted my paper in a PDF file generated by LaTex, the professor said, “All good! Next time just give me an editable version then you are all set.”
I did my Masters in Australia in one of the top unis here. I used LaTeX for a paper and it was flagged for plagarism because the software they used to detect it couldn’t parse some of the characters. It only supported docs from MS Word and Google Docs.
The worse thing is the head professor didn’t understand there was no plagarism looking at my document. I was called in and I had to physically point to the reference I made in the document.
One of the worst experiences I’ve ever had. No idea how these people become teachers.
6 Likes
Re: Benevolent non-compliance
I’m at a bit of a cross-roads myself on my team. My team has been only me for multiple years. I deploy industrial systems and manage them over their entire lifecycle. It has grown to over 15 unique systems across multiple revisions to the overall architecture.
The drive for change could be seen from the outside as NIH syndrome, or a maturation of myself as a software developer. But the theme has been reducing dependencies for a while. (See building systems simply by matklad: https://youtu.be/jVC4DP-8xLM?si=YjK7LiCHBXGSpOpb)
Generation 1:
At first I depended on an IDE that only runs on windows to compile code in a proprietaty programming language that could only run on proprietary hardware. No capability to write tests. No CLI. You literally gotta click the build button.
Generation 2:
Proprietary runtime / IDE, windows only, click to build, no cli, but any Linux preempt rt based hardware, and the language is internationally standardized (structured text). I even reverse engineered the runtime to package it into docker container so I could deploy it more easily from CI. But I’m comitting this click build artifact into the repo manually each deploy. It’s a 50/50 split on what can be tested and what can’t. No plain text source code or tests on the half that’s in the IDE.
Generation 3:
Open source zig package, python control loops, same Linux hardware, all Linux build environment, target is docker on Linux preempt rt, all the software is free. High risk of technical challenges to real time capability and weird shit that just screws you at 9pm on a saturday. But I can write tests that run in CI on Linux. I use WSL only and I only support Linux as the build environment. I depend on about 6 open source libraries across zig and python. Build / deploy takes about 2 minutes with a single cli command.
So I would say my generations have really only been seeking better software engineering principles. I mean I couldn’t even test before.
But the team must grow and I can only hire people that are familiar with Gen 1 / gen 1.5 development. And the hardware I interact with is most easily configured by Gen 1 and Gen 2 (you can click things in the IDE vs becoming expert at reading datasheets). I could only get to managing 15 deployments myself by getting to gen 3, but my team is pressuring a regression to gen 2. Hiring people that don’t write tests (they are only familiar with the gen 1 or gen 1.5 way of working).
For upcoming gen 4, if I prioritize my own happiness:
Integrate zig build system and introduce zig as the first compiled language to the project. Up until this point we have only depended on open source externally managed by me zig precompiled executables. This will enable higher performance and faster builds, and mitigate some of the risks of gen 3. But doesn’t alleviate my personal burden of being the technical lead / primary contributor for everything.
If we go back to gen 2 I do not look forward to setting up a winblows gui enabled ci environment and not testing half my code. Maybe I’ll just leave.
So have I been benevolent? My systems are more performant, I can collect data from them at unprecedented rates and volumes. My tests run on every commit and before every deploy. I can actually test things. My setup.sh script only installs uv and docker (maybe I add zig), compared to other teams 2k line scripts with bazel etc etc etc. But I have raised the bar on who I can hire into my team to a level that makes it hard to hire. I need people that actually write tests and touch hardware, kind of like an embedded software engineer, but those don’t exist or are hard to find.
I kinda wish I could start my own company and deliver / maintain systems the way I want to (system integrator) but customers may require gen 2. But I likely have more resources where I am now :/.
Welcome to life in industrial controls.
I realize this is a bit off topic since the talk was more centered on AI, but I hope to have the environment or resources where I don’t have to swallow a blue or black pill and just regress the knowledge / practices that I have accumulated over the past 4 years. In fact AI adoption is a gen 3 / gen 4 encouragement but it still regresses the reliabillty of my systems under high review load and low technical knowledge accumulated by my team members.
5 Likes
hackernews part is so true
and the user jeffbee pushing back against the “racing 2 dram speeds” is somehow some “new methodology”. he got censored/downvoted.
then dang warned (simlevesque), cause he said he didn’t “trust a word” from anthropic. dang said it was a personal attack. (i got a short video on that too). bunch of jiggery-pokery
Thanks @andrewrk for this talk, it resonated with me a lot. I’m of those who chose to quit. I’m 44, I quit twice, first in 2010 after a PhD in computational neuroscience, and a second time in 2018 after a year as a research engineer. Today I’m doing carpentry for a living.
I left computer science because I felt everyday like I was participating in building a world I didn’t want. And today I’m still doubtful about the emancipatory potential of computer technology - not at the individual level, but on a global scale, when you take into account all that is implied and necessary in order to put that cool device in your hands.
Yet, I’m here this morning hanging on ziggit, watching your talk and reflecting on my path. I still like doing computer stuff, reading about computer stuff, despite my techno-critical views on all this.
I’m not sure where I’m going with this, I guess I needed to write it out (does it makes sense to say that ?)
17 Likes
Is there an alternative video source please? I wont give these BigAd/Ai companies another cent of my attention.
1 Like
i agree with almost all the viewpoints presented in the talk, but i have one critique:
the idea that putin “hacked” the american public to vote for trump.
this is a common view among liberal och somewhat left-leaning people. i dont doubt that the russian goverment tried to meddle in the election process, but i think its dangurous to assume that large parts of the american population didnt make a rational choice and voted in their best interest. i think its tempting to suggest this when someone as divissive as trump gets elected. but its in essence a very condesending viewpoint. you are basically questioning the trump-voters ability to understand and take part in a democracy. speculations along this line can suggest things of elitist and un-democratic nature, like a college professor who, after trump was elected the first time, suggested that voting-rights should be withheld to only them with a college degree or high iq.
so if we assume that the trump-voters wasnt fooled or played to a large degree, what does that mean? we then have to consider that there were and maybe still are real concerns and issues in these peoples lifes that made them vote for him, that they thought he was the most rational choice.
a leftist analysis of this would say: the working class of america have seen their living standard decrease during the last 40 years. this is in direct opposite direction of how tecnology has developed and the wealth of the richest has expanded. and the one mainstream political party that was suppose to represent the the working class interest, the democratic party, has failed to take these issues serously. instead they have focused a lot on issues more relevant for the upper middle class, ie identity politics and the culture war. so abandoned by the political establisment and instututions that suppose to care about them, the working class vote for the candidate that can “shake up the system”. and whatever you think about trump, you have to agree that he has done that.
and you can also see this mirrored in all of europe, with far ringt wing populist stealing votes from formerly social democratic parties.
if this post is wildy off topic, or to political any mods can feel free to remove it.
EDIT: spelling
3 Likes
All Trump has done right was making the gas from 1.2 per litre to 1.7 per litre, and people in Fort McMurray, Alberta finally became happier.
I am serious though…
If every country eventually wakes up and becomes more self resilient, and more focusing on their own energy security, and expending their domestic market, and domestic industry… So they don’t need to make and sell cheap to America, and don’t need to buy America… If these become long term outcomes, Trump is not wrong at this aspect though.
Don’t get me wrong. If Americans eventually want to go back to the production line, and willing take a few dollars per hour pay to make your own Made in America leather shoes, T-shirts, hand bags, underwears, that’s good for you.
(post deleted by author)
Interesting talk. If programmers need a pastor, here he is.
I certainly can relate to the inverse conversion in relation to Christianity. It can be hard to find a new concept of the world and the meaning of life.
Interesting social personal details. Not sure if ‘try to be very likeable for everyone’ is a healthy advise though. Sounds unhealthy and very American to me.
1 Like
I think the point is not that they were acting irrational, but rather that they were making a decision based on false or at least incomplete information. The scary part is that even the most rational person can fall for a good advertisement/disinformation campaign.
2 Likes
I have been seeing myself “taking the black pill” more and more recently, losing hope at the “constantly ceding ground, never gaining it” nature of today’s news cycle that Kelley described. Being one of the (at least according to my own perception) few people still vehemently refusing to use or endorse LLMs remaining, seeing basically the entire software landscape get ensloppified to no end has been absolutely crushing. If there’s really not a computing future without fashware, then, maybe computing is not for me anymore, so I thought.
But the message of the talk, even if a little imperfect, has resonated with me a lot. I’m going to try to keep working in my projects, even if they may never manage to compete with coercionware or even get finished. I think I probably just need more community to not feel so doomer-y about everything, anyhow. Great talk.
10 Likes