#intermediate
Every post tagged "intermediate".
222 posts · showing 141–150
-
Output Went Up, Review Capacity Didn't
PR volume doubled and review capacity didn't, because review is human and got no faster. The number to watch isn't merge throughput — it's how fast approvals are arriving.
-
What Actually Got Automated Away in Previous Software Waves
"Automation always creates more jobs" is too coarse to use. What each previous wave specifically removed — and what it left — is a much sharper predictor, and this wave's profile mostly matches.
-
The Engineer Got Faster and the Team Got Smaller — Both Are Happening
The optimists and the pessimists are both describing real observations from different rooms. One variable decides which room you're in — and it isn't model capability.
-
Cheaper Software Means More Software: The Demand Argument
Every previous drop in the cost of building software increased the number of builders. The mechanism is real and depends on one condition — which you can check in your own organization's backlog.
-
Why 10× Productivity Doesn't Mean One-Tenth the Engineers
The arithmetic is clean and both halves are shaky: the multiplier applies to a fraction of the job, and the conclusion only follows if the amount of software anyone wants is fixed.
-
The Parts of Your Job an Agent Can't See
The code, the tests, the tickets — that sounds like most of the job. What's missing is why the system is shaped this way, what's about to change, and what the requester meant rather than wrote.
-
Raw SQL or Curated Queries? Giving an Agent Your Database
A two-line tool description can hand an agent unbounded access to your data and your query planner. The guardrails matter more than the choice — and one cheap planner call prevents most of the damage.
-
Agents Get Worse While You're Not Touching Them
Nobody edits anything and six months later it handles a whole class of request worse. Five things decay on their own — and none of them announce themselves by breaking.
-
From Prompt Edit to Signal in Under a Minute
A six-minute iteration buys you ten experiments a day; a thirty-second one buys a hundred. Four fixes, about an afternoon of work, and the seam that everything else depends on.
-
How Many Eval Cases Do You Actually Need?
18/20 versus 19/20 is not an improvement — it's noise wearing a number. What your suite size can and cannot detect, and the paired design that makes a hundred cases behave like seven hundred.