Jeff Maxwell (@jeffreymaxwell) header
Jeff Maxwell (@jeffreymaxwell) profile picture

It takes empathy to deliver software with sympathy.

Chicago
Joined April 2009
Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)Photo by Jeff Maxwell (@jeffreymaxwell)
The job: Deliver features to production as fast as possible without limiting your ability to deliver features to production as fast as possible.
2
Jeff Maxwell retweeted
i miss 85
42
61
4
2,423
68,089
Jeff Maxwell retweeted
Python's venv system sucks and the inventor should feel bad, you're telling me every venv has to be updated independently on my machine? I have hundreds.
71
11
15
617
43,379
Jeff Maxwell retweeted
I dunno. My employer still benefits from entire classes of bugs being prevented at compile time, & we’ve had the inevitable “too much AI slop is being committed without review” meetings.
Yes, more and more people are discovering this was just solving puzzles for dopamine, not doing any actual impactful work? 🔥 #clojure
1
2
75
1,902
Jeff Maxwell retweeted
GraphQL exists because getting the backend team to add three fields to an endpoint was harder than inventing a query language.
44
160
26
3,563
125,432
Jeff Maxwell retweeted
"we'll use a NoSQL database so we don't have to worry about schema migrations." - someone who is about to spend the next two years reinventing relational joins in application-level javascript.
116
326
62
6,716
329,684
Why the Midwest is currently my favorite part of the US 1) It still manufactures a lot, and that gives people good meaningful jobs that ground them to reality. The American Dream is still alive here. 2) People are not performative, with a genuineness divorced from promotion. What is the stereotype to play to? Naive, boring, sincerity with a touch of drunk? 3) and like all of the US, every city here has bunch of brew-pubs and great restaurants. The idea that you can only get lattes, home-crafted IPAs, etc, on the coasts is crazy. Everywhere in the country now has great options, especially here. 4) Oh, Great Lakes beaches are vastly under-rated 5) Culver's In praise of the Midwest, my latest piece walkingtheworld.substack.com…
88
136
41
1,432
84,832
Jeff Maxwell retweeted
we named it random access memory (RAM). then we built three levels of cache, prefetchers, data-oriented design, and an entire performance-engineering discipline whose whole purpose is making sure nobody accesses it randomly.
35
80
6
2,357
91,240
This, a thousand times this!
Calculus and analysis is the mathematics that most clearly shows why we should trade “classical” mathematics in for constructivist mathematics: <publish.uwo.ca/~jbell/invita…>
1
2
178
Jeff Maxwell retweeted
“Rust == C++ skill issue” Tell me you don’t know Rust’s origin story without telling me you don’t know Rust’s origin story.
Kinda funny how the comments here are saying it's an undersold advantage; it's to my knowledge pretty much the entire reason @mozilla productionized Rust! Kinda makes me wonder what went wrong there image-wise; maybe "fearless concurrency" transcended from "benefit" to "meme"?
6
1
36
4,736
There are 4 types of people saying "AI will replace Software Engineering" in 6 months 1. they own a company that benefits from replacing engineers (ceo, stockholders etc). they need to justify the amount invested into them 2. Company that made big cuts and they want to attribute it to whatever and AI is good excuse 3. People who never worked as as/with Software Engineering and have no clue the work is not just typing 4. Burned out engineers who don't want to do that job anymore
123
168
19
1,934
119,350
Jeff Maxwell retweeted
🚨BREAKING: Microsoft Research + Salesforce just dropped a paper that should scare every AI builder. They tested 15 top LLMs GPT-4.1, Gemini 2.5 Pro, Claude 3.7 Sonnet, o3, DeepSeek R1, Llama 4 across 200,000+ simulated conversations. Single-turn prompt: 90% performance. Multi-turn conversation: 65% performance. Same model. Same task. Just... talking normally. The culprit isn't intelligence. Aptitude only dropped 15%. Unreliability EXPLODED by 112%. → LLMs answer before you finish explaining (wrong assumptions get baked in permanently) → They fall in love with their first wrong answer and build on it → They forget the middle of your conversation entirely → Longer responses introduce more assumptions = more errors Even reasoning models failed. o3 and DeepSeek R1 performed just as badly. Extra thinking tokens did nothing. Setting temperature to 0? Still broken. The fix right now: give your AI everything upfront in one message instead of back-and-forth. Every benchmark you've seen was tested on single-turn prompts in perfect lab conditions. Real conversations break every model on the market and nobody's talking about it.
Readers added context they thought people might want to know
This is not breaking news, the paper citied was submitted on the 9th of May 2025 and is almost a year old. Link to the paper: arxiv.org/abs/2505.06120
673
1,649
415
8,800
1,630,242
Replying to @GBraggsJr23
@GBraggsJr23 Quick 17-game math vs the 2013 Broncos/Peyton: to match Denver’s 2013 scoring pace (37.9 ppg), the Bears would need 644 points.
1
18
For Caleb to top 2013 Peyton’s 17-game passing pace: 59 passing TDs + 5,820 yards. If someone insists on comparing rounded yards-per-game (e.g., 342.3 vs 342.3), then 5,821 guarantees his per-game average shows up higher even after rounding.
1
11
Completions math: to beat the completions-per-game pace mark (Drew Brees 2016: 471/16 = 29.44), he’d need 501 completions. If he only wants the total completions record, he’d “only” need 491 to pass Brady (2022).
15
Jeff Maxwell retweeted
Finally got here.
2
3
93
Jeff Maxwell retweeted
Honestly, watching Claude Code struggle to figure out how the FUCK the @awscloud CLI parameters and arguments work is incredibly validating.
50
44
11
1,534
116,689
The real Super Bowl is the friends we made along the way.
13
No Wrights? That's wrong.
#Bears have three Pro Bowl selections: LG Joe Thuney (4th in his career) S Kevin Byard (3rd) C Drew Dalman (1st)
22
In retrospect Oliver Sacks having largely made up his “case studies” more as literary exercises than as attempts to accurately recount interactions with his patients shouldn’t be a surprise, since part of their wonder was always how unexpected his findings and interactions were, but I nonetheless find myself surprised… and highly disappointed. The death of “expertise” chalks up another casualty, or perhaps more accurately the dawning realization that as in so many cases the “expertise” was never alive to begin with grows a bit more clear
Bombshell: Oliver Sacks (a humane man & a fine essayist) made up many of the details in his famous case studies, deluding neuroscientists, psychologists, & general readers for decades. The man who mistook his wife for a hat? The autistic twins who generated multi-digit prime numbers? The institutionalized, paralyzed man who tapped out allusions to Rilke? Made up to embellish the stories. Probably also: the aphasic patients who detected lies better than neurologically intact people, including Ronald Reagan's insincerity. newyorker.com/magazine/2025/…
35
64
5
561
25,340
Reminder 1,234 that a great deal of pop science is BS.
Bombshell: Oliver Sacks (a humane man & a fine essayist) made up many of the details in his famous case studies, deluding neuroscientists, psychologists, & general readers for decades. The man who mistook his wife for a hat? The autistic twins who generated multi-digit prime numbers? The institutionalized, paralyzed man who tapped out allusions to Rilke? Made up to embellish the stories. Probably also: the aphasic patients who detected lies better than neurologically intact people, including Ronald Reagan's insincerity. newyorker.com/magazine/2025/…
13
37
1
440
21,839