Latest Try 3

A Facebook engineering research paper (“What It Would Take to Use Mutation Testing in Industry”, arxiv 2010.13464) studied 26 developers using mutation testing in their workflow. The empirical finding: “24 of 26 developers expressed that mutation testing exposed a lack of testing,” and “almost half would actually act on the mutant presented to them by adapting an existing or creating a new test.” What makes mutation testing stick (vs coverage, which gets gamed): a surviving mutant is a concrete, actionable artifact — a specific source-line mutation that passed all your tests silently. Coverage tells you you should write more tests (vague, gameable with assert(true)); mutation testing tells you “this exact mutation passed silently, fix it” (specific, hard to game without genuine assertions).

Latest Try 3

On Medium’s Build with Shopify publication, developer Shahzaib Ali Hassan wrote about why he switched from Liquid themes to headless Shopify with Hydrogen: I initially developed Shopify stores using Liquid templates and themes, but switched to Headless Commerce when clients requested faster, more dynamic, and app-like experiences. The setup uses Shopify Hydrogen and the Storefront API — separating frontend from backend, building the storefront in modern frameworks like React or Next.js while Shopify keeps handling payments / inventory / fulfillment. Pattern: agencies don’t have to choose between all-in on Shopify and ditch Shopify entirely. Headless lets you keep the boring infrastructure (PCI compliance, tax, shipping rails) while taking back the parts the client actually sees and feels — branding, performance, UX. The Partner commission still applies, but the visible-Shopify-branding problem dissolves.

Latest Try 3

On a Show HN post, Jaycobski wrote: I’m a marketer (6 years in SaaS) who spent the last year learning to build software using purely AI assistance (“vibe coding”). I just shipped my first production app for the Oil & Gas industry. The product, BasinCheck, replaces clipboard and Excel workflows for Safety Managers in the Permian Basin — offline audits, hot work permits, automated OSHA 300 logs. A niche B2B vertical that traditional dev teams wouldn’t bother building at startup-scale ROI. Pattern: vibe coding’s success cases cluster in places where (1) the builder has deep domain knowledge in a non-software field and historically couldn’t translate it into code, and (2) the vertical is narrow enough that big SaaS vendors haven’t bothered — so the LLM’s imperfect output is still better than the spreadsheet that currently exists.

Latest Try 3

A widely-cited critique across the 2026 productivity literature: While AI may be accelerating the coding portion of the job, coding represents a relatively small slice of how engineers actually spend their time — planning, alignment, scoping, code review, and handoffs remain largely untouched. Pattern: when you say AI doubled my output, that’s the coding subtask. If coding is 30% of your real workday, a 2x on that subtask is only a ~1.3x on your total throughput. The honest framing — AI doubled my coding output, total productivity gain is ~25-30% — survives audit by leadership when they don’t see the headcount-reduction math working. Same number, very different conversation.

Latest Try 3

Multiple HN discussions on AI coding cite the same shape: the early adopters who plateau around 2 months are the ones who stopped investing in their own workflow infrastructure. The ones who keep getting faster invested in context management artifacts — CLAUDE.md / AGENTS.md, structured task lists, separate planning sessions. Pattern: the addictive easy mode of month 1 is the no structure needed, model figures it out phase. Once you’ve used the tool enough that your tasks involve more state and history than fits in a single chat session, the model stops figuring it out — and you experience that as the model getting worse. The skill that distinguishes month-3 users from month-2 users is willingness to do the boring structural work (write the spec first, break into tasks, version the prompts) that the month-1 honeymoon let you skip.