Update from DEGENBID

DEGENBID weekly update: fewer requests, faster pages, better mobile UX

September 26–October 2, 2026

This week we worked through the layers behind DEGENBID: blockchain data, RPC calls, PostgreSQL, hosting, page loading and the mobile experience. The goal was to make the board easier to use while cutting unnecessary infrastructure work.

Graph / Goldsky

We now fetch new protocol events instead of downloading the full history on every refresh, share identical concurrent requests, and keep requests within the provider's rate limits. In a live quiet-interval check, refresh payload fell from 104.5 KB to 1.6 KB: 98.4% less data, with the same complete event history. We also paused the unused 1.0.0 subgraph while keeping production 1.0.1 live. That removes duplicate worker usage; fewer queries alone do not reduce Goldsky's worker charges.

RPC / Alchemy

We grouped compatible contract reads into Multicall and removed duplicate work. Recovery for 20 payments went from 61 RPC requests to 2; a balance scan went from 29 to 11. ENS lookup for 200 addresses went from 16 to 2 in tests. A live check confirmed that 20 identical simultaneous block reads share one RPC request. These changes preserve fresh reads and payment checks.

Database / Neon

We replaced repeated full-history reads with aggregates and narrower wallet data, reused stable legacy reads, and stopped the legacy keeper from writing on Arena traffic. The 18-hour follow-up recorded zero new keeper-lock updates and zero PocketEntry reads, with 16.37 MB of additional database transfer. Earlier and later traffic differed, so we are tracking usage rather than claiming a fixed percentage saving.

Hosting / Vercel

Multiple timers now share one clock poll. Away from epoch transitions, its interval is 60 seconds instead of 15: 75% fewer requests per clock stream, while the on-screen countdown still updates every second. Hidden tabs stop polling, duplicate display requests are combined, and failures respect backoff and Retry-After. Total hosting cost savings still need a comparable traffic window.

Page speed

We reduced unnecessary startup downloads and automatic route prefetching. Controlled desktop builds measured 22% less initial homepage JavaScript and 34.6% less on the rules page during that optimization pass. The October 1 production mobile Lighthouse audit scored both Holders views at 100, with the basic view downloading about 19% less JavaScript. Homepage performance still needs work; lab scores are not field Core Web Vitals.

SEO and discovery

We shipped robots.txt, a sitemap for nine public pages, page-specific titles and descriptions, canonical URLs, social previews and site-name structured data. The sitemap was submitted to Search Console, and Google's live check confirmed it could fetch it. Indexing and rankings remain up to Google. We also added WebMCP support for reading and searching the board in compatible browsers.

Mobile and onboarding

We rebuilt navigation around a bottom dock, compact team cards and clearer bid actions. Referral and invite pages fit small screens better; chat input no longer triggers unwanted mobile zoom. We also improved chat/profile data preparation and the return to the board after a confirmed team bid.

The rules page is now a playable, wallet-free demo: choose a team, place a practice bid, watch the ranking change and explore your prize share.

Next: keep measuring infrastructure usage, improve homepage mobile loading, and collect feedback on the first-time experience.

Try the practice arena: https://degenbid.fun/rules

Explore the board: https://degenbid.fun