What to read instead of doomscrolling

Everything above came out of one pass. The filter loads a batch of recent public Twitter posts in the niche you picked, reads the text of each one, and sorts it into Read, Skim or Pass. The percentage on a Read is how clear the call was. Open the ones marked Read, glance at the ones marked Skim if the topic matters to you, and leave the greyed ones alone. Then the run ends, which is the part a feed never does.

You open Twitter meaning to check one thing, and forty posts later you are still there with nothing you would repeat to anyone. The posts are not bad, exactly. They are just mostly filler: a hook with no story under it, a screenshot of a quote, a launch, a thread promising ten lessons, an ask for replies dressed up as a question.

That happens because what travels and what is worth reading are two different things. Bait travels because it asks for a reply. A promo travels because someone spent a week sharpening the first line. A platitude travels because it is easy to agree with and costs nothing to repost. Meanwhile the post carrying a real number, a decision that went wrong, or a result someone actually measured sits there with nineteen likes, four posts below the fold.

So the problem is not that you lack willpower. It is that you are doing the sorting yourself, one post at a time, with no way to see what you are about to get before you get it.

Every run starts fresh. It pulls public posts from the last two to three days in one niche, hands each one to Jev, a decision model built by TypeSafe AI, and asks the same five headline questions about every post (Jev answers eight in total). The answers decide the bucket. The whole pass costs a fraction of a cent and takes a few seconds, which the counters above the feed show you as it happens.

Is it specific, meaning does it name a number, an event or a thing that happened. Does it say something new, or is it the same line this niche has read a hundred times. Is it engagement bait, the reply-for-a-DM, comment-below, tag-a-founder kind. Is it a promo, there to sell, launch or hire. Is it a platitude, the sort of sentence that sounds true and tells you nothing. Two of those questions speak for a post and three speak against it, and the side panel shows you the answers for whichever post is under the cursor.

Read means the post says something specific that you probably have not already seen this week. Skim means the answers landed in the middle, usually because the post is short, is a reply out of context, or leans on an image the text never explains, so a glance settles it. Pass means the questions came back the wrong way round, so the card greys out and stays there if you want to check the call. The percentage on a Read is how clear the call was, not a grade for the writer, which is why a 62% Read and a 91% Read are both worth opening.

It judges the text of the post and nothing else. No follower count, no account history, no reputation, which is why two posts from the same person can land in different buckets on the same run. It does not open your timeline, because it has no way to: there is no login and no account involved. It does not tell you whether a post was written with AI, and it does not predict how far a post will travel. It answers one question, which is whether this is worth the thirty seconds it would take you to read.

1. Pick the niche closest to what you actually work on. The chips run along the top of the tool, and the one you pick goes into the page address, so a niche you use often can live in a bookmark.

2. Watch the pass, or do not. It takes a few seconds either way. The feed shrinks as the Pass verdicts land, which is the honest picture of how much of a niche is filler on any given day.

3. Open the Read pile. The list at the bottom is sorted by confidence and every item opens on X in a new tab, so you can middle-click through the three or four you want and read them in order.

4. Close the tab. That is the part that only works because the list has an end. Once a day is plenty: the window is two to three days wide, so a daily run rarely shows you the same post twice, and a weekly run still catches what people are replying to.

Most of the tools in this category are blockers. One Sec makes you wait and breathe before the app opens. Opal and Freedom put the site behind a schedule you set in advance. ScreenZen adds friction at the moment you reach for the icon. They work, and they work best when the problem is the reaching rather than the reading. If you open Twitter without deciding to, start with a blocker. Nothing on this page competes with that.

The second group swaps the feed for other content: a reading app, a podcast queue, a habit tracker that rewards you for staying out. These are pleasant and they solve a different problem. The scroll was never really about the content, so replacing the content only holds if the substitute is genuinely better than what you were getting.

This page is a third kind, and it exists because most people are not going to quit Twitter. Half your work happens there. So keep the feed and remove the filler: one niche, one pass, a short list with a bottom to it. The difference that matters is the ending. A blocker ends your session by force, a replacement app ends it when you get bored, and a filtered list ends when you run out of list. You will know you are done, which is the one thing an infinite feed will never tell you.

The honest list is short, because the useful moves are the plain ones. Put the phone in another room while you work, which beats any app. Move the feed to a laptop so it is not in your pocket at midnight. Replace the reflex with a specific thing rather than a vague intention: one chapter, a walk without headphones, the message you have been meaning to send, twenty minutes on the project you said you would start.

If you are scrolling to feel informed, pick one source with an end to it, a daily paper or a newsletter, and read that instead. If you are scrolling because you are lonely at 11pm, a feed is the worst available answer and one real reply to one real person is the best one. And if you are scrolling for work, because your niche is on Twitter and you cannot ignore it, that is exactly what the filter above is for.

The presets are the niches this was built for first, Jev at the head of the list: Jev, Build in public, AI agents, Indie SaaS, Dev tools, Design, Marketing, Founders, Crypto, Product, Growth, No-code, Open source, Coding agents, LLMs, SEO, Newsletters, Creator economy, E-commerce, Fintech, Sales, Remote work, Productivity, Data and ML, Game dev. Each one has its own search phrases behind it, written by hand, so a run stays inside the topic rather than drifting into whatever was loud that day. The list grows, so if yours is missing, say so and it can be added. Inside SuperX the same filter runs on any topic you type or on your own saved feeds, which is where it goes once presets stop being enough.