Guide · how to read
Every day, fresh science from arXiv becomes text anyone can read. Here's how it all works — before you dive in.
What this is
Every day arXiv — the largest open library of new science — publishes hundreds of papers. We take their full text, hand it to an AI, and ask for one honest thing: explain it clearly. Not shorter, not louder — just clear.
Every article comes out in several languages and at three depths. All free, non-commercial, forever. We adapt a paper only when its license allows it.
who and why
There is one project, but people come here for different reasons. Below is what we have for you personally, what you can take right now, and what we are still building. If you don't find what you need, write to us: we prioritize work based on real requests rather than our own plans.
Just curious
Every day, fresh scientific works retold in simple words. Four levels of depth, from one paragraph to a full analysis, switchable on the fly. Five languages. Semantic search: ask "why is the sky blue" and you'll find an article about light scattering, even if those exact words aren't in it. A 3D map of the project you can fly through. Like and feedback without registration.
in progress More medicine and biology: we currently have a disproportionate amount of physics.
I want to understand and learn
A training course: 14 topics with three paragraphs each, each with an interactive stand, a step-by-step formula breakdown, and a target-hitting challenge. Next to each paragraph is a conversational partner you can ask questions in your own words: "where did this two come from?" It sees exactly this material and doesn't answer with generic phrases. Points and titles are earned for answers, from Curious to Academic. Plus reference books: formulas, constants with units, math explained simply, a retrospective of discoveries.
in progress Personal account login: for now, points live in your browser and aren't available on another device. And more interactive stands — we want every topic to be hands-on, not just a few.
I am a scientist
Semantic search through our analysis of over two thousand works, in any language: ask in Arabic and it will find matches in English. Each article has a link to the original on arXiv: we don't replace the source, we lead you to it. A connection graph "topic ⇄ law ⇄ scientist" you can navigate. Reference lists are parsed — 27 thousand links, and you can see which works our corpus relies on most often.
in progress Researcher's workspace: a place where you can talk to an agent that sees our corpus, upload your data, find similar and contradictory works, discuss experimental design. This is our biggest ongoing construction, and it's not finished yet.
I want to publish my work
Send it to us, it's free. We review works by independent researchers with goodwill, not condescension: we look at novelty, section fit, provide analysis and improvement suggestions. A published work gets a preprint code, and you get a token to withdraw it from publication at any time. Rules, formatting requirements, and a preparation prompt are in the "become an author" section; a work assembled using this prompt is reviewed faster and more accurately.
in progress We currently handle submissions semi-manually: the response won't be instant, but it will definitely come. We'll automate this later — first we want to ensure the conversation with the author feels human.
I teach · educational institution
Everything is free, no registration, no ads, and no data collection about your students. Material in five languages, including Arabic with full right-to-left layout. The course, reference books, and articles can be used as is — via a link, without any agreements with us.
in progress We are ready to customize selection for your curriculum, add sources if their license allows, and add a language. We'll honestly tell you upfront how far our resources can go. For very specific tasks, your participation in development under open licenses may be needed.
I write about science · I translate
Take it. Our texts are under CC BY 4.0 license — you only need a link to the source. Headlines, covers, figures, and graphs from original works with captions — all in place. Four levels of depth are handy when you need to quickly grasp the essence before your own material.
in progress If you spot an error — tell us. That's the most helpful thing you can do for us: we correct it and note what exactly was fixed.
Partnership
The project's structure allows building a similar one for a different task: your own sources, language, selection, audience. Write to us, let's discuss without obligations.
in progress Honest about limits: the team is tiny, resources are modest, we don't take on everything and not right away. But we say this upfront, not after you've waited a month.
I want to help
The most valuable thing is a constructive comment: what's unclear, what's broken, where we got it wrong, what's missing. Write to any of our addresses, we read everything.
in progress The project has a small circle that influences its direction. You can get in not with money or by the number of pages read. For now, this opportunity exists.
The core idea
Not "short" and "full", but a ladder. Start wherever you're comfortable and go deeper if it grabs you. The depth switch sits at the top of every article.
Simple simple
For a first encounter.
When the topic is new and you want to grasp the essence without jargon. Any jargon here is always spelled out.
Popular popular
Once science already draws you in.
A lively account with intuition and imagery. The default level — every article opens here.
Advanced advanced
With formulas and the story of the discovery.
Closest to the original. For those who want depth and aren't afraid of math.
Mini mini
Just the gist in 10 seconds.
One thought, like a post in a feed. To decide whether to read on.
Site map
Left to right — from the feed to a living map of knowledge. The logo always brings you home.
A feed of articles by day. An endless stream grouped by date — like a news feed for science.
Concepts. A single knowledge registry: from black holes to methods and instruments. Each concept has a class (phenomenon, method, constant, unit…), a full card with tabs — Description · History · How it works · In practice — and every related article. On top of classes there are sections by meaning: statistics — the data methods papers use without naming them; constants — with value and unit taken from our formula anatomy and the SI definitions; mathematics — the apparatus the laws are written in.
Laws and principles. The "house of formulas" — laws, theorems, effects. A formula is only a rendering; the meaning is in the text.
Formulas. A cloud of formulas from the advanced write-ups. Each one comes with a symbol-by-symbol breakdown: what every variable means, constants with values and units, operators. Plus history, limits of applicability, and the articles where it works.
Scientists. The great minds behind the discoveries — with a biography and related topics.
arXiv sections. Cosmology, exoplanets, high-energy physics… — a filter by field.
Authors. Researchers publishing on arXiv. An A–Z index; each author has a page with all our write-ups of their work.
The knowledge graph. Tags ⇄ laws ⇄ scientists and all their links. You can wander through knowledge, not just read it. There is also the concept graph: frames instead of the whole cloud (groups → one group → neighbors), link power = shared articles, class shapes, 2D/3D, ring and sphere views, kind and group filters.
The course. A knowledge tree: 14 topics of three paragraphs each, every one with an interactive stand, a worked-through formula and a hit-the-target challenge. Alongside it: a retrospective of discoveries, a reference of formulas and constants, maths made plain, and bold hypotheses.
Project ideas — things you can take on: real work with methods, a first step and an honest risk. Every idea rests on named papers; where our archive is silent, the arXiv field stands in, and that is marked.
Favorites. Articles you've marked with a heart, gathered in one place.
Languages
The language switch is in the row under the menu. Russian is the source language; the rest are prepared by AI.
How to search
The 🔍 button in the header searches everything at once: articles, concepts, laws, scientists and sections — just type a word. The prefixes remain for those used to them, and now merely narrow the view:
For example, #black_hole — articles on the topic, @Hawking — by author, !Einstein — linked to a scientist. Next to it, the 📅 Calendar: year → month → day — jump back to any date. And the sections filter narrows the feed to the arXiv fields you want.
Search has two modes. By words — instant and offline: matches titles and concepts. By meaning — the query goes to the server and is compared against the meaning of the articles themselves: “why is the sky blue” finds the article on light scattering even if those words never appear in it. Meaning search works across all languages at once: ask in Arabic and it will find English articles, answering with Arabic titles. While it thinks, a fact or a short question from our own articles appears instead of empty space.
If something is unclear
In the course every paragraph comes with someone to ask: put it in your own words — “where does that factor of two come from?”, “what is this for in practice?” — and get an answer about this very material, not a generic one. The bot sees the paragraph text and the formulas right beside you.
Answers to quizzes and challenges earn points, and points add up to ranks — from Curious to Academician. The ladder is borrowed from the real one: observer, lab assistant, graduate, PhD, doctor of science. For now it all lives in your browser; once sign-in exists, the score can be carried over.
The special part
Articles, concepts, formulas and scientists form one graph. A concept here is a topic, a law, a quantity and a constant alike — each with its own card. Links come from three sources: appearing together in articles, closeness in meaning, or a stated relation (“part of”, “describes”, “measured in”). Every concept page carries a live mini-map of its surroundings; concepts/graph holds the whole map — an overview of areas, one area in full, or the neighbourhood of a single concept.
To the author of the work
At the bottom of the full analysis, there is a section addressed not to the reader, but to the author of the article and their colleagues. We do not review others' work—that's what peer review is for, and we don't do that. We show the surrounding field: what, in our reading, has been done in the work, what its strengths are, which neighboring works in the archive lie nearby, and where one could go from here.
Neighbors are found not by word matches, but by meaning: each article has its vector, and nearby are works that say the same thing in different words. Each direction must be based on a specific work from our archive—it is named as a link. No basis, no direction: a forced suggestion devalues the real ones.
What to explore
A separate page — “What we propose to explore”. These are questions you can take on: each has a plan of a few sentences (where to start, what to measure, how to verify), a list of papers worth reading, and an honest survey of the field — how many arXiv papers lie close to the topic and how many of them we have analyzed.
Map and analytics
Beyond individual articles there is a bird’s-eye view: the project map — a 3D cloud of everything we’ve covered. A dot is an article, nearby dots share topics, colour marks a topic group the algorithm found on its own. Group names and descriptions are written by AI, in plain words.
The map has four views: Articles — the cloud of works; Authors — every researcher we processed, along an “experimentalist ↔ theorist” axis; Fly — fly through the cloud and open articles like stars; Heat — a 3D landscape of “topic × month”.
The Dashboard holds the numbers: articles per day and month, which fields of science are covered, what people read, plus how many papers exist in the open arXiv archive and how much of it we have covered so far.
Nice things to know along the way
express and full
Some articles are marked “express” — a quick take built from the author’s abstract only, without parsing the full paper. Full articles are written from the whole text: context, methods, results, formulas — deeper and more detailed. Express can be hidden from the menu, leaving only full ones.
where this comes from
The source is arXiv, the largest open preprint archive: works appear there before journals. Every day we take what came out the day before yesterday, not yesterday: arXiv adds works with a delay, so "yesterday" turns out to be incomplete. That's three hundred to a thousand works per day.
From these, twenty to twenty-five are selected. There are two conditions, both mandatory: the work must be useful to science and interesting to a person without background. Emphasis on natural sciences. We only take machine learning and mathematics where they serve natural science or our own work with scientific texts, not for their own sake: otherwise the feed would turn into a stream of neural network articles — simply because arXiv today is a quarter composed of them.
Why the vector is here. Each work is converted into a set of numbers reflecting its meaning — this is how the machine compares articles without understanding words. This underpins semantic search, finding similar works, and connections between topics. The model is cross-lingual: an Arabic query and an English article land in the same place, so language ceases to be a barrier. Now we are also training the vector to select articles of the day — on par with the language model, not after it.
we listen
Under an article, in the course, after a challenge — everywhere you can mark "like" or "dislike" and write a few words. Without registration and without a name. This is not decoration: feedback shows which retellings turned out unclear, which explanations don't work, and where we missed the mark with the level. Silence is also a signal, but a bad one: you can't tell what to fix from it.
your own paper
You can send your paper directly to article@bridge42worlds.academy. We check it with AI: whether it is novel and whether it fits a section. If it does, the work may be published here in our processed form. Please format it to the usual standards. We reply either way, and it is free.
We do not limit ourselves to particular sources. Coursework by university or school students is welcome too. Only two things matter: the work must meet the standards expected of scientific writing, and it must relate to physics or engineering.
cadence and feedback
For now we cover the top 10 papers a day, with a focus on physics. Next comes the full stream, unfiltered.
The project can be adapted to a concrete need: connect new sources where their licence allows it, add a language, tune the selection to a given field. Proposals go to proposal@bridge42worlds.academy.
Cooperation with educational institutions is entirely free. We are glad to take your wishes and requirements into account: add new sources, or adapt the feed to your programme, as far as our resources allow. Very specific tasks may also call for your own help with the development, under open-source licences.
About the name. bridge42worlds reads as bridge for two worlds — the world of science and the world of a person outside it. A bridge is genuinely needed there: a paper clear to ten specialists stays invisible to everyone else. The digits standing in for “for two” are a quiet nod to The Hitchhiker's Guide to the Galaxy, nothing more. We do not have forty-two worlds. There is one world, and it is ours.
Want to become an author? We review work by independent researchers — honestly, kindly and for free. The whole path, the preparation prompt and the rules: become an author →
Would you like to take part in the project? Write to admin@bridge42worlds.academy.
The project has an observers' council — a small circle of readers who decide where it goes. We do not invite by page count: invitations go to those whose remarks are visibly to the point, and first of all to authors of the papers we covered. If that speaks to you — write to us, the door will be found.
Need more information? Write to info@bridge42worlds.academy.
Liked it, or didn't — tell us: author@bridge42worlds.academy
about money
There are no ads, paid tiers, or donation collections here, and we don't plan to introduce them. The project runs on modest expenses and the time of those who make it. If you ever feel like supporting it with money — you'll find a way without our hint, and we won't give one. A constructive comment is more valuable to us, and that's not a figure of speech but a calculation: we can find money, but understanding what to fix is hard to come by.
We keep this page up to date: it describes what works today, not what is planned. If something here differs from what you see on the site — the discrepancy is our fault, and it's worth writing to us about it.