BooksCentral

What is in it

Catalogue

Book and author pages that members create and correct themselves, with every edit passing through a moderation queue, full revision history, and a route for authors to claim their own page.

Reading life

Virtual shelves, ratings, reviews, author follows, friends, recommendations to friends, an activity feed and notifications — the social layer that turns a catalogue into somewhere a reader returns.

Rankings

Weekly BookMARK and AuthorMARK charts, computed in each country's own timezone from page views, shelf adds, reviews and follows, with a Bayesian prior so a handful of votes cannot top the chart.

Bestseller lists

Country editors upload each week’s bestseller list as CSV, and entries are matched to existing catalogue pages by ISBN or by title, so the list links through to real book pages.

Revenue surfaces

Affiliate buy links with click tracking, an Author Pro subscription, a reader Premium subscription, and sponsored placements with an advertising fallback — all present from the first release.

What does running one country at a time actually mean?

Running one country at a time means each country gets its own catalogue, its own weekly bestseller list, its own rankings and its own address, such as pk.bookscentral.org. A single account works across all of them, so a reader signed in on one country site is already signed in on the rest.

The alternative — one global catalogue with a country filter — fails on the thing readers care about most, which is relevance. A bestseller list that mixes markets is nobody's bestseller list, and a ranking dominated by the largest market makes every smaller country a rounding error. Separating them means a Pakistani reader sees Pakistani rankings, computed in Pakistani time.

The usual cost of that separation is operational: a new country becomes a deployment project. BooksCentral avoids that by choosing the country database from the hostname before any application code runs, so opening a country is a subdomain, a database and one admin form rather than a release.

Why do the weekly rankings need a Bayesian prior?

The weekly rankings need a Bayesian prior because raw averages reward obscurity. A title with a handful of five-star ratings from friends would otherwise outrank a title with hundreds of ratings and a strong average, and the chart would stop meaning anything within a week of launch.

The prior pulls every score toward the population mean in proportion to how little evidence stands behind it, so a title has to earn sustained attention rather than a burst of it. Rankings draw on page views, shelf adds, reviews and follows rather than one signal, because any single signal is trivially gamed, and they are computed in each country's own timezone so a week means the reader's week.

A ranking’s credibility is invisible when it works and obvious the moment it fails. A chart that a small coordinated group can capture loses reader trust immediately, and trust is the only asset a ranking has.

What does BooksCentral demonstrate for a client project?

BooksCentral demonstrates multi-tenant SaaS engineering that many firms describe and few have shipped: a separate database per tenant selected at the edge of the request, single sign-on across subdomains, and an interface that meets WCAG AA and mirrors for right-to-left languages.

The tenancy model is the transferable part. One codebase serves every country, one identity database holds accounts and billing, and middleware picks the country database from the hostname before any handler runs. That shape applies directly to any product sold per region, per franchise or per white-label partner, and it is far cheaper to design in at the start than to retrofit once a second tenant exists.

The rest is ordinary discipline made visible: 157 backend tests run against both MySQL and MariaDB, a matching 157-test frontend suite, 13 end-to-end browser journeys, and automated accessibility checks at four screen widths. VenSoc builds client systems the same way — BooksCentral is simply the one where the entire codebase is ours to show.