Alright, listen up. Intel just dropped. 🕶️ Early Bird tickets for RubyConf go on sale March 16th — and they are LIMITED. Mark the date, stay sharp, and be ready to move. We only get one shot at this.
Check out rubyconf.org for news & updates!
The RubyConf CFP closes next week on March 15! Get those submissions in!
Same from last post: The journey of learning has all types of twists and turns, gaps, uncertainty, and it's a struggle.
Same as previous post: Lemur with straight path from A to B, what we think learning is.
Hrm, transparent backgrounds on dark mode with a dark line don't work well, let's try that again
Learning illustrated as a straight line between A and B, lemur running towards finish line.
The actual path of learning, curves, double backs, cliffs, skips, the lemur holding on for dear life rather than having a clean path to finishing.
I'd brought up a similar point in a talk a while back: Learning is the journey, not the destination, and the path to learning is never straightforward. Skipping that deprives someone of all the instinct building they get along the way, which is where seniority is forged.
So basically what he kept doing to Square / Block
We’re preparing to announce our first RubyConf 2026 keynote. This year’s keynote speakers represent voices that push the language and community forward in meaningful ways. Stay tuned for the reveal. 🫢
It's been a while, but sometimes when you first read something you're not really ready for it
Square is not the company I left anymore. The level of callousness to fire half the workforce is a tragedy. I'm so sorry to anyone impacted by this change and hope you all land safely.
Yeah. That and curating them for repeat use. I swear I spend so much time trying to get a pallete going that I admittedly sometimes use Kuler or AI to get ideas and dial things in.
When I'm making a talk it's all the repetitive elements like shading that get to me across tens of illustrations. That and curating and deciding on swatches.
True enough, there's always _some_ room to do things but the idea of orienting on capabilities is that those lower the barrier to effectiveness from an order of months to weeks to potentially days or hours.
The cheaper we make it to do the right thing the more likely it is to happen organically.
The job now becomes acting as a catalytic force to fundamentally change how teams operate, and the level they can operate at.
We're not staring at blank sheets of paper wondering what to do anymore, we're refining on top of an imperfect vision of what could be.
That's a dramatic shift.
It still does in many ways, but now we have ways to condense that into tools which make reasonable approximations of that expertise, enough so to get them above the threshold of justifying further investment.
It does not need to be 100% accurate, it just needs to prompt investment.
That's still true, but a majority of that work was around capability. We need to raise the base level of engineers to where they are capable of this work, with or without seasoned experts present.
Previously that required a lot of political capital to up-level and train.
How do we increase the capabilities of teams? How do we programmatically up-level them and make this work cheaper and easier to deliver? How do we make sure that they understand the problem / framework / language?
I've said in the past that monolith decomposition efforts are a very social problem.
Normally my posture in that situation is to raise a campaign and get teams to buy in with a deliverable date far enough out that they can close existing threads, but that takes months and a lot of coordination to land.
With AI I've found myself very heavily weighting towards #3.
Using AI I have taken work that should have taken multiple months and delivered it within a week. The work was not difficult, but the coordination was, and the capacity was just not there. That meant that I needed to fundamentally change the equation to land things.
Often times #2 is extremely difficult, so we get into heroics which captures #1, but not every company rewards that. Some will, some will actually punish it.
The needle that ICs without HC/Planning can move tends to be more around #3:
How do we make this work cheap, easy, and fast?
With any large scale tech debt project there are three things needed to succeed:
1. Commendation - Anyone who works on this is incentivized, recognized, and rewarded.
2. Capacity - Teams have dedicated resources to making this happen.
3. Capability - Tools, resources, and knowledge exist
This is our fate. For years we mocked it, derided it, and now? It shall finally have its' true revenge.
Clippy has returned.
Keep yourselves safe. I wish I could tell you it's safe and you all should come anyways, but it's not and I'm not sure I'd risk it if I were in a similar situation.
That was our reaction too when he bought it and I was still working there.
Never 😛
Mostly remembering the days of Chef.
.....you sure that's not also Rust?
SysAdmin meant Perl,
DevOps meant Ruby,
SRE meant Go,
What are we at nowadays? Infrastructure Engineer and Python?
I, uh, might still have a few Perl books.
Ah IRC: xkcd.com/1782/
What're folks hosting it on now? Freenode imploded and I stopped paying attention to it for a while.
The more I work with Ruby the more I am inclined to turn off autoloading and explicitly require things. Really, the more I'm inclined in general to be explicit about things.
What is simple in the early days becomes a tax of trying to trace down what and where something is.
What if new hires could understand your entire codebase on day one?
@baweaver.bsky.social thinks this is possible with MCP tools.
An AI agent could become an expert in your codebase and make institutional knowledge available 24/7 for any team member.
Watch his SF Ruby Conference talk:
I still have mine if you ever want to borrow it.
