Steve Yegge: Well, so I have just started to play with Claude 4, which has been out for a week, but I’m a little behind, because of the book and everything… So what I wanted to do is extrapolate from Claude 3.7’s performance and Claude 4’s performance, and see if I could make any right projections about where I think it’s going to be by the end of the year… Because I have a bunch of – here’s the thing, is I have a bunch of tests that I can give it. Obviously, they’re not like comprehensive eval suites like a company would run, but I’ve been struggling with a set of problems that Claude Sonnet 3.7, which is the best coder out there, hasn’t been able to get past. Real simple stuff like client-server RPCs just really seem to confuse it. Or anytime you’re going over some sort of network boundary and you’re trying to make it make changes on the client and the server, it was just outside of its sort of cognitive bounds. And you could always see it. You could try different prompting. It went on for weeks, and I just found that I had – I had a problem that was slightly too big for it. And I have probably seven of these now, and I’ve been giving them to Claude 4. I gave my first one last night, and it just – bang, it just did it. It’s like “Hmm… Okay.” So it’s a definite, definite increase. I don’t know how much, if it’s 20% or 100% yet… But what I can tell you is that the way they’re increasing, and from experts like Jason Clinton, who’s CSO at Anthropic… He told us in April at the ITRev Forum that AI has been getting 4X increases in cognitive power for decades. That’s been following – it’s a function of Moore’s law. And the experts, the consensus in the AI community is that there are probably – there are at least two more cycles left in that progression before something changes. It either slows down, or AI finds a way to speed it up. But either way, it’s going to be 16 times smarter than it is today, in June/mid-2028. And what does that even mean? What does it mean for it to be 16 times smarter? Does that mean it goes from 10 IQ to 160 IQ? I mean, I don’t I don’t know what it means, but it definitely means they’re going to be smarter than us. And so, I got a chance to meet Dario… Did I tell you guys about that?

Jonathan Carter: One year. One term is one year. But I want to return to technical stuff. I’ve so many ideas, and so many – well, relatively small projects, but that has a big impact. Next year I want to focus on that. I have so many ideas to make the installer stuff better, to make it easy to find the [unintelligible 01:16:28.11] One person called our downloads page an IQ test, because just finding the Debian image they want to difficult for them.

Paul Smith: Well, absolutely. I mean, absolutely. Like, just for myself… And I don’t think that this is a prerequisite for somebody who believes that government should work – government as a function of something that we do collectively together… You don’t have to believe that President Obama was a good president, or that you work on his campaign as a prerequisite to have worked on the healthcare.gov rescue. That was an important aspect for me, but I will say that we were hearing all the time – so the Affordable Care Act was already the law. Healthcare.gov was sort of the delivery mechanism. But we were hearing all the time from people for whom the law had already made their lives better. They could stay on their parents’ health insurance longer, until they were 26. Or they couldn’t be denied coverage because of a pre-existing condition. Those stories were filtering up to the White House, and then down into the team. So it’s visceral, right? It’s people’s lives, and you have this almost direct connection to them. And so yeah, it gives you – like, when you’re flagging a little bit, like, you’ve worked all day, it’s nine o’clock on a Saturday, and you’d rather just be done… It gave us all that extra little bit of “Well, we can’t really slack off here. We have to take this over the finish line.” So yeah, that was definitely an important part.

I wanted to go back and say that the team that we’ve encountered - so we were talking a little bit about how there was some team that built it, and this is a combination of government contractors and government agency folks. Our mission and our belief as a team was to have high EQ first; bring our high IQ about web stack, but we weren’t there to blow anything up. There were like six or seven of us at the beginning; we weren’t going to rewrite healthcare.gov in a couple of weeks, or whatever. We needed them to succeed. So really, what we did more than anything was point the way to “Here’s what this thing should be doing. Here’s how you’re on a path to success incrementally”, by adding monitoring, by having a process by which we sort of prioritize bugs, and defects, and tackle them in sort of a reverse order of their impact, and here’s what the sort of indicators of a successful high-traffic website look like, and how we can move closer in that direction.

So really, our innovation, if anything, on the rescue itself was bringing – one of our team members was Mikey Dickerson, who was a site reliability engineer at Google. One of the early people at Google who kind of helped create that culture - bringing some of those ideas to government, so having a daily standup where all thestakeholders could talk about their technical issues, and we could coordinate and communicate and prioritize and plan… Which - none of that was happening before we showed up on the scene. So it created a sense of urgency, it created accountability, which is good… And not just like finger-pointing or blame, but “Hey, we really need you to do this thing. And it’s really important, because we need this bug to be fixed”, or whatever. And people really rallied to that. So we wrote very little code, although we did write some Go code, that turned out to be pretty load-bearing.

Jerod Santo: By the way, I looked up that Intel x86 hack… Meltdown and Spectre. Now it probably rings a bell. Spectre and Meltdown, if you all recall those. All the best, all the most non obvious high IQ hacks have cool names. There’s no cool name for the one where you forgot to not send the token back on the password reset form. That one doesn’t get a name, but it happens probably way more often.

Alright, let’s move forward here… Number seven, “Monorepos are easier to audit.” That one seems like – I don’t know, that one seems obvious when you think about it, because like, well, everything’s in one place. But is there more to it than just that?

Nicholas Mohnacky: So that’s what I mean by vectorizing and bundling your IQ and your knowledge; and potentially – like, you’re a walking, talking digital wallet of Chris’ intellectual property that you could contribute somewhere else, or just get paid on for everything that you’ve already contributed.

Mat Ryer: It sounds like the president doing an IQ test.

Paul Smith: I mean, absolutely… Just for myself – and I don’t think that this is a prerequisite for somebody who believes the government should work… Government as a function of something that we do collectively together. You don’t have to believe that President Obama was a good president, or that you worked on his campaign as a prerequisite to have worked on the HealthCare.gov rescue. That was an important aspect for me, but I will say that we were hearing all the time – so the Affordable Care Act was already the law; HealthCare.gov was sort of the delivery mechanism.

But we are hearing all the time from people for whom the law had already made their lives better. They could stay on their parents’ health insurance longer, until they were 26, or they couldn’t be denied coverage because of a pre-existing condition. Those stories were filtering up to the White House, and then down into the team…

So it’s visceral. It’s people’s lives, and you have this almost direct connection to them… So yeah, when you’re flagging a little bit, it’s like – you know, you’ve worked all day, it’s nine o’clock on a Saturday and you’d rather just be done… It gave us all that extra little bit of like “Well, we can’t really slack off here. We have to take this over the finish line.” So yeah, that was definitely an important part.

I wanted to go back and say that the team that we encountered – so we were talking a little bit about how there was some team that built it, and this is a combination of government contractors and government agency folks… Our mission and our belief as a team was to have high EQ first… Bring our high IQ about the web stack, but we weren’t there to blow anything up. There was six or seven of us at the beginning - we weren’t gonna rewrite HealthCare.gov in a couple of weeks, or whatever. We needed them to succeed.

So really, what we did more than anything was point the way to “Here’s what this thing should be doing. Here’s how you know you’re on a path to success incrementally, by adding monitoring, by having a process by which we sort of prioritize bugs and defects, and tackle them in sort of a reverse order of their impact… And here’s what the indicators of a successful, high-traffic website look like, and how we can move closer in that direction.”

So really, our innovation, if anything, on the rescue itself was bringing – one of our team members was Mikey Dickerson, who was a site reliability engineer at Google… One of the early people at Google who kind of helped create that culture… Bringing some of those ideas to government. So having a daily stand-up where all the stakeholders could talk about the technical issues, and we could coordinate and communicate and prioritize and plan, which - none of that was happening before we showed up on the scene.

So it created a sense of urgency, it created accountability, which was good… And not just like finger-pointing or blame, but “Hey, we really need you to do this thing, and it’s really important, because we need this bug to be fixed”, or whatever. And people really rallied to that. So we wrote very little code, although we did write some Go code, that turned out to be pretty load-bearing.

Adam Stacoviak: I know we’re gonna talk about that at some point more deeply, but is that underlying what that is? Like, once you learn coping skills and have awareness, and you have a real clear picture of what you’re actually thinking or why you’re thinking it and all these things - is that emotional intelligence, to some degree? Is that what that is, like building up – not an IQ…