A while back I posted here about RealFeedApp, a free digest that collects science and tech news, collapses the retellings into one card, ranks by significance and writes a short summary in your language.
It worked. I stopped drowning. And then I noticed I was doing something odd with it: I'd read a summary, understand every word of it, and still have no idea whether the thing mattered.
What a summary doesn't do
A summary compresses an article. It cannot supply the thing the article assumed you already had.
Every piece of technical writing is addressed to someone. A paper on gene therapy is written for people who know what a vector is; a post about attention layers is written for people who have already met attention. The author skips the setup, and they're right to — their reader has it.
Compress that article faithfully and you get a shorter text with the same hole in it. Here's the shape of it, with the details changed:
Researchers report that the modified vector achieved durable expression in the target tissue without the immune response seen in earlier trials.
Every word true, nothing wasted. And if that isn't your field, you've learned nothing you can act on. Is this a lab curiosity or the thing the field has been stuck on for a decade? The summary can't say, because the article never said — the article was talking to someone who knew.
This is the actual cost of reading widely. Not volume, which is what everyone builds for. Volume is solved by filtering. Context is not, because the missing part was never in the text you filtered.
What's new: three lines under the summary
So every card now carries a second block, under the summary and behind a divider. Three labelled rows, no heading:
Field — a short tag: how neural networks work, gene therapy, hardware and chips, energy and batteries.
What was done — the concrete thing, in ordinary words.
Why it matters — what changes if it holds up.
Same three rows on every card, in the same order, in your language. The point isn't more text. The point is that the same three questions get answered every time, so you can skim the third line across ten cards and stop at the one that deserves you.
A few decisions inside it that changed how it reads:
The field is a tag, not a sentence. It comes from a fixed list of areas, so the same subject is worded identically on every card. That's what makes it scannable — and it's why it can become a filter later. Free-form text would have been easier to build and useless the first time the same field got named two different ways.
No heading above the block. It's three labelled rows; the labels already say what it is. A heading would have eaten the space the explanation needed.
It doesn't appear where there's nothing to explain. If the linked page had no article on it — a video, a page that's really an app, a PDF that yields nothing — there's no block. There's also no card: pages with nothing to read stopped reaching the feed at all, which fixed a smaller embarrassment I'd been shipping for a while. Summaries of "this page requires JavaScript" are, it turns out, very fluent.
That last rule matters more than the feature. An explanation of nothing is worse than no explanation, because it looks exactly like the real thing and you can't tell without opening the link — which is the one thing the card exists to save you.
What it doesn't do
It doesn't read the article for you. Three lines are enough to decide, not enough to know, and the link to the original is still the whole point of the card.
In your own field it will occasionally explain something you learned a decade ago. That's the trade: the block is written for the reader who wandered in from three fields over, and I'd rather it be redundant for the specialist than absent for everyone else.
Try it
Still free. Open the link and you get a preview of the feed with no account at all; signing in with Google opens the rest of it and remembers your topics. realfeedapp.com
The Android app is in closed testing; it'll reach Google Play once I'm confident it behaves.
Two things I'd like to hear if you open it:
- Does the third line earn its space? "Why it matters" is the row I'm least sure about — it's the one where an honest we don't know yet would be more useful than a confident sentence, and I don't know yet whether it's doing that.
- Where does the field tag fit badly? The list of areas is fixed, so somewhere there's a story it can only be wrong about. I'd like to know which one before a reader has to.
Top comments (0)