A reading list is the easiest thing on the internet to publish. Twenty links, a sentence under each, no position taken anywhere, and there’s nothing in it anybody can accuse you of being wrong about.
The version worth reading is the one where things got cut, so there’s a list near the bottom of what I stopped opening. Everything above it is a dozen sources I still read, grouped roughly by what each one is for.
Two of them I found because somebody put them in a list like this one.
The recruiters have their own trade press
Hung Lee, weeklyRecruiting BrainfoodHand-curated recruiting news, written for recruiters. Read one issue and job ads stop looking like descriptions of jobs.Hung Lee has been putting out a weekly, hand-curated digest for the talent business for years. He calls it “the industry newsletter for the talent business”, and the word doing the work there is industry. You’re reading somebody else’s mail.
A job posting is the last artifact of a long internal process, and by the time it reaches you almost everything interesting has already happened. The headcount fight is over. Somebody’s internal candidate got floated and either survived or didn’t. The agency contract is signed or it isn’t. The recruiter who ends up with your application is being measured on something specific this quarter — time to fill, submit-to-interview ratio, a hiring manager who keeps rejecting the shortlist for reasons nobody will put in writing. You can’t see any of that from where you’re standing, and it’s all discussed openly in the trade press, which exists to help those people do that job and talks about their problems the way people talk when they assume the affected parties aren’t in the room.
You won’t use any of it directly. What changes is what a posting looks like to you, and how personally you take a silence.
Recruiters guess right about 55 percent of the time
Almost everything people repeat about resumes and screening is folklore. It gets passed hand to hand, never checked, and it usually started with somebody who last looked for work in a different decade. There is real data. Most of it comes from one company that ran experiments on its own hiring platform and then published the results, including the ones that made the platform look bad.
The study to read first: with the economist Peter Bergman, interviewing.io collected nearly 2,200 evaluations from 76 technical recruiters across more than 1,000 resumes, then checked those predictions against how the candidates actually did in the interviews. The recruiters were right 55 percent of the time. Two recruiters looking at the same resume differed in their pass predictions by an average of 41 percentage points. Median time spent per resume: 31 seconds.
Aline Lerner and Peter Bergman, interviewing.ioAre recruiters better than a coin flip at judging resumes?The study behind the 55 percent. The disagreement numbers matter more than the headline.The 41 points is the figure that should stay with you. Two qualified people at the same company, looking at the same page, routinely reach opposite conclusions about you. So a single rejection tells you almost nothing, and rewriting your resume on the strength of one is superstition. Ten applications with no phone screens is a different matter. That’s when you rewrite.
The same shop ran blind technical interviews for six years and found that 46 percent of the people who landed offers from top-tier companies through it had neither a top school nor a top company on their resume. The paper screen was removing people who could do the job.
interviewing.ioFor six years, we ran the largest blind eng hiring experiment of all timeWhat happens when you strip the resume out of the process entirely. The 46 percent figure lives here.interviewing.ioStop trying to make recruiters thinkWritten against a median of 30 seconds of attention. Specific, unsentimental, and it’ll make you delete things.One columnist, whatever you do for a living
Alison Green has been answering workplace letters at Ask a Manager since 2007. She gets about 50 letters a weekday, and the site takes roughly two million views a month. It’s the only thing on this list I’d hand to a nurse, a paralegal, and a staff engineer without changing a word.
What makes it good is that she answers the question that was asked, then answers the question underneath it, and she’s willing to tell a letter writer that the thing they want to do is a bad idea. Nearly twenty years of that has produced an archive covering almost every awkward moment a job search can generate: chasing an offer that’s gone quiet, resigning without burning the reference, being asked for a number first, getting ghosted after a final round.
The comment section is its own weather system and I go back and forth on it. Some weeks somebody with the exact job in question turns up and the thread is better than the answer. Some weeks it’s the whole room relitigating one word in the letter. I keep reading it anyway.
The way to use the site is as a lookup table. Before you send anything uncomfortable, search the archive for your situation. Somebody has already asked, and the answer will be more specific than anything a search engine hands you.
Three for engineers, and I’d stop at three
Gergely OroszThe Pragmatic Engineer“Big Tech and startups, from the inside.” Gergely runs a recurring tech jobs market series — the 2026 installment on hiring managers and job seekers is where I’d start.Will LarsonStaffEngStories and guides from people who reached Staff without managing. Worth reading before you accept a title you haven’t defined.Yangshun Tay, MIT-licensedTech Interview HandbookA structured prep path that costs nothing, with 140,000-plus GitHub stars behind it. Enough people pay for a worse version that it’s worth saying plainly.One more, and this one is a single post. Julia Evans wrote Get your work recognized: write a brag document in 2019, and it has quietly done more for the people I’ve sent it to than any interview guide I know of.
You don’t have to try to make your work sound better than it is. Just make it sound exactly as good as it is!
A brag document ends up being the raw material for every resume bullet and interview story you’ll need eighteen months from now, captured while you still remember the numbers.
Every comp number you’ll find was typed in by somebody who felt fine about it
Crowdsourced compensation datalevels.fyiSelf-reported offers by company, level, and location. The best free starting point, with the caveats below.Everything on levels.fyi was typed in by a person who chose to type it in, which tilts the whole dataset toward people pleased with their offer. So work in bands. What you want out of an hour there is “senior at this size of company in this city clears X.” Cross-check against Blind, which is anonymous, frequently sour, and occasionally the only place a specific band gets said out loud. It’s where I go to learn that a number exists at all.
The other half is being somewhere a stranger can find you
Reading is the passive half of this. The people who hand you a job are rarely your close friends, because your close friends already know about the same openings you do.
Rands in ReposeRands Leadership SlackGetting on for 40,000 engineers and engineering leaders. You email to join and a human reads it, which is most of why the signal survived.LumaLuma — Discover EventsIncreasingly the default listing for tech meetups and small conferences. Filter by your city and go to something with under 50 attendees.Hacker News, monthlyAsk HN: Who is hiring?A monthly thread that reliably draws several hundred postings, written by hiring managers rather than sourcers.The Hacker News thread deserves a specific instruction. Open the current month, use your browser’s find function for “REMOTE” plus your stack, and notice how many entries include an email address belonging to somebody who will actually read it. There’s a sibling thread — “Who wants to be hired?” — where you post yourself instead. Both run every month, and both are among the few remaining places where a first contact isn’t mediated by an applicant tracking system.
I checked whether these are still alive or merely still scheduled. July’s hiring thread ran to 436 comments and the wants-to-be-hired thread to 566, which is in line with the months on either side of it. What I can’t tell you is the conversion rate. Comment counts are the only public number, and they don’t say how many of those threads ended in a job.
What I stopped opening
- Anything with a number in the title promising resume tips. The interviewing.io study above is what a real answer looks like, and producing it took a controlled experiment and an economist.
- Newsletters that summarize other newsletters. You’re two layers from the source and the second layer didn’t add anything.
- LinkedIn commentary about hiring being broken. It’s true, it’s widely agreed on, and it has never once told anybody what to do on a Tuesday morning.
- Communities whose pitch is access to hiring managers. When the introduction is the product, you’re paying for an introduction from a stranger, which is worth about what it sounds like.
- General career podcasts. I gave four of them a real try and got roughly one actionable paragraph per hour of audio, and reading a paragraph takes a minute.
Forty minutes a week, or it doesn’t survive
A list like this fails the same way every time. You subscribe to all of it in one enthusiastic evening, and six weeks later you haven’t opened any of it.
The version that survives is smaller than you want it to be.
- Subscribe to two. Whichever two map to the job you’re trying to do this month — trade press and data if you’re still working out why nobody replies, comp sources if you have an offer coming.
- Give it one slot on a calendar, forty minutes. Reading that happens in the gaps doesn’t happen.
- Read toward a destination. Every session should end with something written down: a company name, a person to email, a number, a sentence worth stealing for your resume. A session that produces nothing puts that source on notice.
- Kill anything you’ve skipped three weeks running. It isn’t going to get better and you already know it.
Step three is the one that fails, and it fails quietly — you read something good, you mean to act on it, and it evaporates into a browser tab you close on Friday. Whatever you use to catch that matters more than which two newsletters you picked. Vigil does the catching for the inbound half, the recruiter mail you didn’t ask for. A text file with a date at the top handles the rest.
Start with Brainfood. Open the most recent issue and read it the way you’d read a trade journal about yourself — written by the people who decide, for the people who decide, with no idea you’re looking over their shoulder. If it makes you uncomfortable, you’re reading it correctly.