Been building tools for researchers, students, and journalists for a while now, and one pattern keeps showing up: everyone's browser is stuffed with extensions they installed once and never trust again.

Every extension you add is more permissions, more background tracking, more RAM eaten, more "why is my browser slow" moments. For people whose entire job is reading and synthesizing information all day, that adds up fast.

A few things I've learned building in this space:

1. Paywalls aren't usually about IP theft prevention — they're about casual readers. Most publishers have open-access mirrors, cached versions, or public archive partners. You don't need to "hack" anything, you just need a faster path to what's already public.

2. Text-to-speech is criminally underused for research. Reading dense academic or legal text is slow. Listening to it while doing something else (commuting, exercising, scanning a second source) roughly doubles your effective research throughput.

3. Summarization tools are only as good as the trust of their source. Generic AI summaries hallucinate details. Pulling from established educational directories instead of freeform generation gives you a much more reliable "first pass" before you dive into the primary source.

The bookmarklet model (vs. browser extension) trades a little bit of convenience for a lot of trust — no background permissions, nothing running when you're not using it, nothing to "clear cache" beyond a button click.

If you're doing heavy research work and your browser has become extension soup, it might be worth simplifying. Happy to answer questions on how any of this works.