<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn" -->

---
title: The cycle of overcorrection | daily.dev
description: Engineering teams tend to overcorrect when reacting to pressure, instead of making measured adjustments. The author, an engineering manager at HoneyBook, gives...
canonical: https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: The cycle of overcorrection | daily.dev
og:description: Engineering teams tend to overcorrect when reacting to pressure, instead of making measured adjustments. The author, an engineering manager at HoneyBook, gives...
og:url: https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn
og:image: https://api.daily.dev/og/posts/LKqiQQ1tN.png
og:image:alt: The cycle of overcorrection
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# The cycle of overcorrection

**[Manager.dev](https://daily.dev/sources/zaidesanton)** · 7 min read · 2 upvotes · 0 comments

## Summary

Engineering teams tend to overcorrect when reacting to pressure, instead of making measured adjustments. The author, an engineering manager at HoneyBook, gives four examples: CI pipelines that went from minimal to a 30-minute flaky suite, AI-code-review rules that swung from 'trust nothing' to 'trust everything', AI adoption that swung from token leaderboards to strict spend caps, and the industry-wide 'flattening' of management layers that Meta later reversed by re-hiring managers. HoneyBook's own AI spend averages $1,500 to $2,000 per engineer. The fix proposed: experiment in small increments and plan for reversing any decision before announcing it.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.manager.dev/newsletter/the-cycle-of-overcorrection>

## Questions this post answers

### How much do companies typically spend per engineer on AI coding tools like Claude or Copilot?

One engineering team reports an average AI spend of $1,500 to $2,000 per engineer, with no per-person cap or token leaderboard. Spend naturally varies by project: engineers on greenfield work use more AI, while some of the strongest engineers on sensitive code use relatively little, and that is considered acceptable rather than a problem.

_Teams weighing AI budget policy can track real-world spend benchmarks like this on daily.dev._

### Should engineers be required to fully understand every line of code written by AI coding agents like Claude?

Forcing engineers to fully verify every line an AI agent writes slows teams down significantly, as they spend extra time questioning and going deep on all types of work. A better approach varies the verification level by task: moving faster with less understanding in some areas, and verifying critically in others, adjusting the ratio over time rather than applying one blanket rule.

_Developers balancing AI-assisted coding speed with code ownership can follow this debate on daily.dev._

### Why did Meta start rehiring managers after cutting middle management layers?

Meta's push to flatten its organization by cutting middle management layers did not work out well, so the company began re-hiring managers. This reflects a broader pattern where companies cut management layers under pressure to reduce costs or speed up decisions, then reverse course once the downsides, like worse communication flow or bottlenecks, become apparent.

_Engineering leaders deciding on management ratios can watch outcomes like Meta's play out on daily.dev._

## Similar posts on daily.dev

- [An Engineering Manager’s crazy year](https://daily.dev/posts/an-engineering-manager-s-crazy-year-qdeatqbeh) · Manager.dev · 22 upvotes · 3 comments
- [The AI Bill Is the Cheap Part — Aviv Ben-Yosef](https://daily.dev/posts/the-ai-bill-is-the-cheap-part-aviv-ben-yosef-msswsjbz0) · Tech Lead Digest · 4 upvotes · 0 comments
- [The Cost of Proof Has Collapsed](https://daily.dev/posts/the-cost-of-proof-has-collapsed-nflsmuiln) · Medium · 0 upvotes · 0 comments
- [Your error budgets don’t know AI exists](https://daily.dev/posts/your-error-budgets-don-t-know-ai-exists-rnyvl6lwu) · LeadDev · 0 upvotes · 0 comments

---

Tags: [#cicd](https://daily.dev/tags/cicd), [#leadership](https://daily.dev/tags/leadership), [#company-culture](https://daily.dev/tags/company-culture)

[View this post on daily.dev](https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"The cycle of overcorrection","url":"https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn"},"datePublished":"2026-11-17T07:16:33.766Z","dateModified":"2026-11-17T07:17:24.184Z","description":"Engineering teams tend to overcorrect when reacting to pressure, instead of making measured adjustments. The author, an engineering manager at HoneyBook, gives...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/becbd6e925d1e4ed29639bd6660e2532?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/becbd6e925d1e4ed29639bd6660e2532?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Manager.dev","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Manager.dev","logo":"https://media.daily.dev/image/upload/s--nwW90hoG--/f_auto/v1731601654/logos/Logo","url":"https://daily.dev/sources/zaidesanton"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"cicd,leadership,company-culture","timeRequired":"PT7M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Manager.dev","item":"https://daily.dev/sources/zaidesanton"},{"@type":"ListItem","position":3,"name":"The cycle of overcorrection"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-cycle-of-overcorrection-lkqiqq1tn#faq","mainEntity":[{"@type":"Question","name":"How much do companies typically spend per engineer on AI coding tools like Claude or Copilot?","acceptedAnswer":{"@type":"Answer","text":"One engineering team reports an average AI spend of $1,500 to $2,000 per engineer, with no per-person cap or token leaderboard. Spend naturally varies by project: engineers on greenfield work use more AI, while some of the strongest engineers on sensitive code use relatively little, and that is considered acceptable rather than a problem. Teams weighing AI budget policy can track real-world spend benchmarks like this on daily.dev."}},{"@type":"Question","name":"Should engineers be required to fully understand every line of code written by AI coding agents like Claude?","acceptedAnswer":{"@type":"Answer","text":"Forcing engineers to fully verify every line an AI agent writes slows teams down significantly, as they spend extra time questioning and going deep on all types of work. A better approach varies the verification level by task: moving faster with less understanding in some areas, and verifying critically in others, adjusting the ratio over time rather than applying one blanket rule. Developers balancing AI-assisted coding speed with code ownership can follow this debate on daily.dev."}},{"@type":"Question","name":"Why did Meta start rehiring managers after cutting middle management layers?","acceptedAnswer":{"@type":"Answer","text":"Meta's push to flatten its organization by cutting middle management layers did not work out well, so the company began re-hiring managers. This reflects a broader pattern where companies cut management layers under pressure to reduce costs or speed up decisions, then reverse course once the downsides, like worse communication flow or bottlenecks, become apparent. Engineering leaders deciding on management ratios can watch outcomes like Meta's play out on daily.dev."}}]}
```

