Starting a blog doesn’t cost money. At least that’s how I started back in the 2000s, and free platforms are still around. Intermediary platforms came along in the 2010s, giving some control back to bloggers or sharing the ad revenue. The way out of it all, ever since the term weblog was coined, was to run it yourself — either on a machine from a paid hosting service or on your own home server. The question was price and effort.

Running it yourself in the early days, early HTML days, couldn’t promise the reliable power and network a hosting service provider could. Soon after, the question of privacy of the personal homepages was raised; mailing address and home IP were publicly available. Yet another problem a hosting solved. The prevalence of bots and DDoS attacks on personal blogs didn’t make it easier to bring the site back home either. And the blogs themselves became more complex — WordPress made publishing easier to manage while demanding more system resource to run. Paid hosting answered all of these needs, so long as one was willing to foot the bill. In exchange, these services took the jobs off of your back.

In my particular case, I wasn’t using the full suite the hosting company was offering. Backup, for instance, was a paid add-on, and there are free battle-tested backup guides and plugins for WordPress, so I built my own routine. That same job likely throttled my hosting plan. Which left me two options: a. take half-baked and longer interval automated solution, or b. just pay for it. I chose the automation over paying for yet another add-on.

I had been paying around $80 per year for hosting, domain, and other add-ons. As is the case with most registrar and hosting services, the prices generally only go up and new paid add-ons are added. The quoted price does not give a good price trajectory for the future. The only time the competitive market reins them in is when a paid add-on becomes the new market standard, such as WHOIS privacy — masking the mailing address that would have otherwise been exposed. However, getting a better plan often requires transferring out of an existing plan, a complete migration.

When the opportunity came, or rather when frustration built to the point of reconsidering, I moved to Cloudflare — even though its free tiers came with all these minor clauses and tags attached. Writing up a migration plan obviously means working within given constraints; if the budget won’t budge, something else has to give. That was the role of the AI in the planning phase. Cloudflare Registrar having no markup was one thing, but the caveat GPT Codex found (the nameserver requirement) meant there are still some levels of compromises I should be willing to make. Albeit it didn’t affect me negatively, it could have complicated the planning phase far worse.

The problematic backup solution, once I migrated entirely into my linux box, was no longer needed. I am running a linux Docker box as a home server, and I already have a working backup model for all Docker stacks. WordPress being well documented for Docker environment was a lucky break. Had it not been, I would have leaned on AI far heavier for a taller plan. What’s left is roughly $10 a year — no more hosting subscription to renew, only a domain subscription with a better deal stayed.