HN Debrief

Self hosted email continues to steeply decline

  • Infrastructure
  • Privacy
  • Open Source
  • Developer Tools
  • AI

The RIPE Labs post looked at MX records across the Tranco top 1 million domains and argued that email keeps consolidating into two providers, Google and Microsoft, while self-hosted mail keeps losing share. It is measuring where domains route mail, not inbox counts, and several readers immediately called out that this says much more about popular domains than about the full universe of personal or hobby mail servers. Even with that caveat, the broad picture landed: email is still an open protocol on paper, but in practice deliverability power is concentrating around a few operators with the reputation systems to get mail accepted.

If email matters to your business, treat deliverability and provider concentration as strategic infrastructure risk, not just an admin choice. Small-volume self-hosting still works for many people, but new setups should assume IP reputation, DNS hygiene, and fallback relays are the make-or-break variables.

Discussion mood

Frustrated and skeptical. People broadly buy the centralization trend, but they distrust the article’s framing, dislike its apparent LLM-written style, and see the real problem as opaque deliverability rules at Google and Microsoft rather than self-hosting becoming technically impossible.

Key insights

  1. 01

    The dataset misses the personal long tail

    The RIPE numbers describe popular domains, not email as a whole. Looking only at the Tranco top million biases the result toward businesses and high-traffic properties, while many self-hosted domains are obscure by design and never appear in that sample. One reader also noted that even within the article’s own chart, self-hosted mail still exceeds any single named provider, which makes the headline feel sharper than the underlying result.

    Read this as a story about concentration among prominent domains, not proof that independent mail hosting is disappearing overall. If you use third-party data in your own strategy work, check the sampling frame before turning a trend into a market-wide claim.

      Attribution:
    • 1vuio0pswjnm7 #1
    • layer8 #1
  2. 02

    IP reputation dominates technical correctness

    Getting SPF, DKIM, DMARC, PTR, and reverse DNS right is table stakes. It still does not buy inbox placement. Several operators said the real determinant is whether your sending IP and surrounding range already have a good history with Gmail and Outlook. That makes migration surprisingly risky and explains why old domains on long-lived infrastructure often work while fresh VPS deployments hit invisible walls despite perfect configuration.

    When planning mail changes, treat IP history as a dependency you need to test before cutover. For a new server, validate reputation early, keep rollback options, and avoid assuming standards compliance will be enough.

      Attribution:
    • spl757 #1
    • mrighele #1
    • groomlake #1
    • yrro #1
  3. 03

    The winning pattern is split inbound and outbound

    A lot of people are no longer choosing between pure self-hosting and pure outsourcing. They keep their own domain, aliases, mailbox logic, or inbound server, then hand outbound delivery to Amazon SES, Fastmail, Proton, or another relay with established reputation. That sidesteps the silent-dropping problem without giving up all control. It is less ideologically pure, but it matches where email’s power has actually centralized.

    If you want ownership without constant deliverability fights, design for hybrid mail from the start. Keep identity and routing under your control, then buy reputation for outbound sending.

      Attribution:
    • groomlake #1
    • cephi #1
    • cjs_ac #1
    • vachina #1
    • gommm #1
  4. 04

    Open protocols did not stop centralization

    Email is a clean counterexample to the idea that interoperability alone prevents platform lock-in. The protocol is open and federated, yet anti-spam enforcement, reputation scoring, and abuse handling have concentrated enough power in a few providers that everyone else must work around their rules. Some readers went further and said this is why messaging has drifted into walled gardens and why a better protocol would need spam resistance and authentication built in from the start.

    Do not assume protocol openness will preserve market openness. If you are building on an open standard, look hard at where trust, abuse control, and ranking power can recentralize the ecosystem anyway.

      Attribution:
    • friendzis #1
    • MoreQARespect #1
    • dbingham #1
  5. 05

    Bulk-mail tactics have crept into normal operations

    The operational advice sounded more like growth email than personal correspondence. People recommended seed accounts across major providers, open-tracking pixels, A/B testing, feedback loops, Google Postmaster Tools, and warming IPs or domains. That is telling. The modern deliverability playbook is no longer just for newsletters and marketing teams. Even legitimate small senders end up borrowing the same machinery because inbox placement has become adversarial.

    If your company depends on outbound email, staff it like a deliverability function, not a basic sysadmin chore. Silent failure is common enough that monitoring across major providers is worth the overhead.

      Attribution:
    • spl757 #1 #2
  6. 06

    Software got easier but maturity still matters

    Newer tools like Stalwart, Mox, Mail-in-a-Box, and NixOS-simple-mailserver have clearly lowered setup friction. But one operator testing Stalwart for business use ran into self-update crashes, weak spam filtering, and flaky 2FA, while others stuck with Postfix because it is boring and predictable. The progress is real, just not enough to erase the value of old, stable mail software when the service carries business risk.

    Prefer proven mail stacks over newer all-in-one convenience unless you have time to absorb instability. Ease of setup is useful, but boring reliability still wins for production email.

      Attribution:
    • lazyweb #1
    • Scarjit #1
    • VladVladikoff #1
    • raddan #1

Against the grain

  1. 01

    Low-volume self-hosting still works fine

    Several long-time operators said the standard horror story is overstated for personal and small-business use. They report years of smooth delivery, only rare blocklist incidents, and migrations that were much easier than expected. The catch is scope. They are not sending bulk mail, and they accept that occasional reputation hiccups are part of the deal. That undercuts the blanket claim that self-hosting is now impractical.

    If your volume is low and your needs are simple, self-hosting is still a reasonable option. Just decide upfront whether you can tolerate rare silent failures and occasional delisting work.

      Attribution:
    • Leonard_of_Q #1
    • yrro #1
    • ryandrake #1
    • sshprotocol #1
  2. 02

    LLM writing was not the main problem

    A few readers pushed back on the pile-on about machine-written prose. They argued that cheap writeups for middling-value topics are economically rational, that many workplace reports are worse, and that some LLM-assisted writing is perfectly readable or even easier to process. That does not rescue this specific article for everyone, but it does challenge the idea that AI-assisted text is automatically disqualifying.

    Judge low-cost technical writing by whether it conveys something useful, not only by whether you can hear the model in the prose. If you publish with AI help, the bar is still on you to edit enough that readers trust the substance.

      Attribution:
    • supermatt #1
    • Arnt #1

In plain english

2FA ↩
Two-Factor Authentication, a login method that requires a second proof of identity in addition to a password.
A/B testing ↩
A method where different users are shown different versions of a product or configuration to compare outcomes.
Amazon SES ↩
Amazon Simple Email Service, a cloud service for sending transactional and bulk email with Amazon-managed infrastructure.
DKIM ↩
DomainKeys Identified Mail, an email authentication method that helps receiving systems verify that a message was authorized by the sending domain.
DMARC ↩
Domain-based Message Authentication, Reporting, and Conformance, an email authentication policy used in DNS.
Dovecot ↩
A widely used mail server component for mailbox access protocols like IMAP and POP3.
Fastmail ↩
A commercial email hosting provider focused on paid email service rather than advertising.
Google Postmaster Tools ↩
A Google service that gives senders data about domain reputation and deliverability to Gmail recipients.
LLM ↩
Large Language Model, a machine learning model trained to generate and analyze human-like text.
Mail-in-a-Box ↩
A packaged self-hosting project that automates setup of a personal mail server and related services.
Mox ↩
A modern email server project aimed at simpler self-hosted mail setup.
NixOS-simple-mailserver ↩
A NixOS module that automates much of the setup for running a mail server.
Postfix ↩
A widely used mail server program for sending and routing email.
RIPE Labs ↩
A publishing platform from RIPE, a European internet registry, where researchers and practitioners post analysis about internet infrastructure.
SPF ↩
Sender Policy Framework, a DNS-based rule that says which servers are allowed to send email for a domain.
Stalwart ↩
A newer all-in-one email server project that bundles multiple mail functions into one system.
VPS ↩
Virtual Private Server, a rented virtual machine used as a server on the internet.

Reference links

Email software and self-hosting tools

Deliverability and anti-spam references

Standards and protocol background

Privacy and identity context

  • Keyoxide
    Mentioned as inspiration for a decentralized identity layer that could reduce dependence on one provider
  • PRISM program
    Cited in a privacy-focused argument against concentrating email with United States providers
  • FISA Amendments Act of 2008
    Linked to explain the legal framework behind data access concerns raised about major providers

AI and content quality references