Wow, that’s an egregious chart crime. Surely there’s a way to make a chart that isn’t so dishonest. Just plotting the SPR level would make the same point honestly.
Hmm in my experience (I've done a lot of head-to-heads), Opus 4.6 is a weaker reviewer than GPT 5.4 xhigh. 5.4 xhigh gives very deep, very high-signal reviews and catches serious bugs much more reliably. I think it's possible you're observing Opus 4.6's higher baseline acceptance rate instead of GPT 5.4's higher implementation quality bar.
I find Opus 4.5 very, very strong at matching the prevailing conventions/idioms/abstractions in a large, established codebase. But I guess I'm quite sensitive to this kind of thing so I explicitly ask Opus 4.5 to read adjacent code which is perhaps why it does it so well. All it takes is a sentence or two, though.
tried replicating w/ a slightly different system prompt w/ sonnet-4.5 and got some different results, esp w/ progressive to conservative questions. Prompting seems pretty load-bearing here
AI tools have been so good for me for making home-cooked software. As a new-ish parent, it’s so much easier to do stuff. I don’t need to go into extra-deep focus mode to learn how to center a div for the hundredth time, I can spend that precious focus time on the problems that matter / the core motivation.
I had a good laugh at the image showing that 2018 had a “quality focus”
I think Jonathan Blow gave his “Preventing the Collapse of Civilization” talk (much stronger treatment of the subject matter) around that time, also about how software quality was only going down
this is not true unless you’re doing some kind of adjustment. For t bills, us03m yields were much higher 30-40 years ago.
> US equities are at all time highs
this is true
> USD falling day over day, month over month, year over year
if falling means inflation, yes in banal way. If falling means relative to other currencies, that’s the last 9 months or so. Previously the USD was quite strong
> US debt is falling in value because no one wants to buy it
this appears to be the hinge of the argument? It is not true. 10y yields have been down / flat since beginning of 2025 (i.e., price up). also tsy auctions remain well-subscribed / within historical range
Just got it at work today and it’s a dramatic step change beyond Cursor despite using the same foundation models. Very surprising! There was a task a month ago where AI assistance was a big net negative. Did the same thing today w/ Claude Code in 20ish minutes. And for <$10 in API usage!
Much less context babysitting too. Claude code is really good at finding the things it needs and adding them to its context. I find Cursor’s agent mode ceases to be useful at a task time horizon of 3-5 minutes but Claude Code can chug away for 10+ minutes and make meaningful progress without getting stuck in loops.
Again, all very surprising given that I use sonnet 4 w/ cursor + sometimes Gemini 2.5 pro. Claude Code is just so good with tools and not getting stuck.
“Redfin data shows millennials are more or less on track with previous generations”
“more or less” is doing some heavy lifting. The chart referenced shows a pretty dramatic discontinuity between Millennials and Gen X / Boomers from age 25 to 35 w.r.t homeownership rate.
Ha! I was thinking of Google specifically. I imagine it varies significantly then. Maybe 40+ L5 is not an OOM more common than 7, but L5 + L6 I think safely is. Agreed on L7 pay, very doable.
I think you’re right, I see that his resume links to Distinguished Engineer roles at Google / Amazon. Which … I don’t know. At my FAANG-adjacent company, there have only ever been _low_ single-digit number of ICs at that level. We’re talking 0.1-0.3% of all engineers. And they had insane track records.
And FWIW I think that there’s at least an order of magnitude more “happy L5s” older than 40 at FAANGs than senior staff+
I’m not sure where these “per day” benchmarks are coming from -— is this supposed to be executive pay or mid-level/senior engineer pay? Because $5k - $10k / day works out to $1m - $3m / yr (depending on if you use 200 working days / yr or just 365). Which, yes, happens (esp with good year of stock appreciation) but is not as common as the prose makes it seem.
Also these numbers come from companies like this? “These companies aren’t Google or Apple, but rather some tractor company or heavy manufacturing company just churning out results for year.” Seems unlikely! The post says they fly under the radar, but are there any examples? In general, non-tech companies pay software engineers significantly worse bc you’re a cost center
And this footnote: “if you do the math using practical inflation and cost of living going up 7% to 13% per year” — if you’re going to claim extraordinary inflation over the last decade like that, please share how you arrived at the number!
5 years ago I did a similar analysis of price per part because I remember there being a big controversy around the price of the new Star Destroyer set. I analyzed price per piece but also price per gram, estimated in a few different ways. I was surprised to see that price per gram was stable through the 2010s on an inflation-adjusted basis and that it went down fairly significantly over the 1990s.
I did this before I learned how to finish projects, so it's been sitting in a private repo the whole time :)