What Changed: Overload Before and After Generation Became Free
Information overload is an old complaint, older than the web, and for decades the standard advice held up reasonably well: skim more, subscribe carefully, trust the front page. That advice assumed something that stopped being true, namely that most of what reached you was written by a person who meant it.
The old overload was a volume problem. Blogs, papers, newsletters, forums: the pile was tall, but nearly everything in it cleared a basic bar, because writing took effort and effort filtered out most of what was not worth saying. You could answer a volume problem with throughput. Read faster, skim harder, let an editor or an algorithm rank the pile, and you got a workable slice of a mostly honest whole.
Then generation became free. When anyone can produce a thousand plausible articles before lunch, effort stops filtering anything, and the pile changes character. The new overload is not too much of a good thing. It is too much of an indistinguishable thing: text that is grammatical, confident, search-optimized, and empty, sitting in the same results page as the real thing and wearing the same clothes. The word for it, slop, moved from niche complaint to household vocabulary fast. In December 2025 it was named Word of the Year by both Macquarie Dictionary and the American Dialect Society, and dictionaries are lagging indicators. By the time they canonize a word, the experience behind it is already ordinary.
This is why the two standard remedies have stopped working. Reading faster fails because speed is a response to volume, and volume is no longer the problem; skimming slop faster just means being fooled more efficiently. Better feeds fail for a more structural reason: the feed is the distribution channel slop was built for. A ranking system tuned for engagement has no inherent preference for human-made over generated, because generated content can be tuned for engagement more precisely than any human writer would tolerate. Asking the feed to save you from slop is asking the ocean to save you from salt.
The Two Scarce Skills: A Filter and a Memory
When production becomes infinite, scarcity does not disappear. It moves. The scarce work used to be writing; now it is reading well, and reading well in 2026 decomposes into exactly two skills, one at the front door and one at the back.
The filter: deciding what gets in
The first skill is choosing sources instead of accepting streams. Researchers and commentators have started describing a kind of attentional fatigue particular to this moment: readers spending a growing share of their effort not on understanding material but on assessing whether it deserves understanding, checking for the tells of synthetic filler before committing attention. That verification tax is charged on every unvetted item, which is precisely why the widely argued antidote in current discourse is human curation, deliberately chosen sources and personally kept collections, over algorithmic feeds. In practice the filter is three swaps. People over algorithms: follow named humans with reputations to lose, not recommendation engines with quotas to fill. Subscriptions over feeds: a newsletter or an RSS folder is a list you wrote, while a feed is a list something else keeps rewriting for you. Deliberate search over scroll: arrive with a question and leave with an answer, rather than arriving with time and leaving with less of it.
The memory: keeping what proved valuable
The second skill is less discussed and, quietly, more important. Suppose your filter works and something genuinely good reaches you. If you do not keep it, you have merely rented it, and the rent has gone up. Re-finding anything on the open web now means wading through the slop a second time, except worse: the good article you remember is now surrounded by paraphrases of itself, generated after it, ranking beside it, each one confidently wrong in ways you would have to read carefully to catch. The judgment you exercised on first reading, this one is real, this one is worth it, is real work, and letting the result evaporate means paying for the same verification twice. A memory turns that one-time judgment into a permanent asset. The filter decides what deserves your attention. The memory makes sure you never have to decide the same thing again.
Most people already sense this, which is why they have improvised memories everywhere: forty open tabs, a bookmarks bar last organized years ago, links messaged to themselves, screenshots taken as a promise to return. These are all attempts at keeping, and they all fail the same way, at retrieval. A tab is not a memory, it is an obligation with a countdown; a bookmark folder is a place things go to be forgotten with a clear conscience. What the moment actually calls for is a memory that costs nothing to feed and stays searchable as it grows, because a memory you cannot query is just a second pile.
The Case for a Personal Library
Put the filter and the memory together, run them for a while, and something with real structural value emerges: a search index where every single entry was placed there by a human who read it first. There is exactly one index like that available to you, and you are the only one who can build it.
The practical habit this enables is simple to state: search your own library before you search the open web. On the open web, every query is a gamble against an adversary that produces plausible text faster than anyone can debunk it. In your library, the gamble is already settled, because everything in it is something you once chose on purpose. The article that survived your skepticism, the PDF you actually read, the note you wrote because something finally made sense. No public index can promise that every result was vetted by someone whose judgment you trust completely. Your library can, because the someone is you.
- The verification tax is paid once You judged the item when you saved it. Every later retrieval inherits that judgment for free, while the open web charges you full price on every visit.
- Retrieval becomes trustworthy A hit in your library is not a candidate to be evaluated. It is a known-good source coming back, often with your own context still attached to it.
- Old material appreciates On the open web, a good page from three years ago is now buried under imitations. In your library, it is exactly where your past self left it, unchanged and undiluted.
Consider the ordinary case. You need to make a decision you half remember reading something smart about, a year ago, maybe two. The open-web version of that moment is a query, a page of results you must audit one by one, and the creeping suspicion that the thing you remember has been shouldered off the first page by imitations of it. The library version is the same query against a few thousand items, every one pre-vetted, with the exact piece coming back along with the note you attached when it first mattered. Same memory, same need, entirely different cost. Multiply that by every question you will ask over the next decade and the case for the library stops being aesthetic and becomes arithmetic.
And the arrangement compounds. Each year of curation makes your library answer a larger share of your questions, which means each year you need the open web a little less, which means the web's ongoing decay costs you a little less. The open web becomes a place you visit deliberately, the way you visit an archive or a market: with a purpose, briefly, bringing back only what passes inspection. Your library becomes the corner of the internet that is all signal, and it is the only corner whose signal rate you control.
An Honest Distinction: AI That Produces vs AI That Organizes
At this point a fair objection arrives. If AI-generated content is the pollution, is it not strange to let AI anywhere near the solution? The objection deserves a straight answer, and the straight answer is that the line does not run between artificial and human. It runs between whose intent the system serves.
AI as a producer generates material and then needs you to consume it. Its economics point one way: text is created because creating it is nearly free and your attention, captured at scale, is worth money. A model that writes feed content is optimizing for engagement, which is to say it is optimizing you, and it succeeds precisely when you read something you did not set out to read. That is the machinery of slop, and no amount of quality improvement changes its direction, because the direction is the point.
AI as an organizer points the other way. When a model tags, summarizes, transcribes, and links material that you already chose, it produces nothing for you to consume and makes no claim on your attention. It is doing librarian work on a collection whose every item passed through your judgment first. The choosing, the only part where taste and trust matter, stays entirely human. What gets automated is the filing, which is the part nobody ever wanted to do by hand anyway. The first kind of AI replaces your judgment. The second kind amplifies it.
The question to ask of any AI in your reading life is not whether it is artificial. It is whether it decides what you see, or takes care of what you decided.
A Practical System for the Slop Era
None of this requires heroics, a productivity philosophy, or a weekend of setup. It is five habits, most of which subtract effort rather than add it.
- Choose a small set of trusted sources A dozen or two named people and publications whose work you have verified over time. Add slowly, on evidence. This list is your filter, and a short list you trust beats a long list you skim.
- Save at the moment of value, with one shortcut The instant something proves worth keeping, capture it and close it. One keystroke, not a browser sagging under forty open tabs, each one a decision you are postponing. An open tab is a to-do; a saved item is an asset.
- Let the library organize itself Do not build folders and do not hand-write tags. Manual filing is a tax that eventually makes you quit, and in 2026 it is a tax software can pay for you. Your job is choosing; the filing is not your job anymore.
- Search your library first Make it the default reflex before any open-web search. Every question your library answers is a trip through the slop you did not take, and the reflex strengthens as the library grows.
- Prune rarely, but honestly Once or twice a year, delete what you saved aspirationally rather than truthfully. A library's value is its signal rate, and honest pruning is how you keep it near total. Rarely, because pruning is judgment work; honestly, because keeping everything is just overload with a delay.
The change arrives quietly after a couple of months. You notice that your first search of the day was your own library, that it answered, and that you never opened the feed at all. Overwhelm, it turns out, was never really about how much exists. It was about how much you felt obliged to face without a filter and without a memory, and you have both now.
Where Mindly Fits
The filter is yours alone, and no app can choose your sources for you. Mindly was built to be the other half: the memory, made so cheap to maintain that keeping it stops being a discipline and becomes a reflex.
Capture is one shortcut. Press ⌘M and the article, PDF, screenshot, file, voice memo, or half-formed note you are looking at lands in your library, and the librarian work you should never do by hand simply happens: the AI tags it by topic, summarizes it if it is long, transcribes audio, reads the text inside images, and links it to the related things you saved before. Later, you search the whole collection in plain language, by meaning rather than by the exact words you half remember, and because everything it returns is something you once chose, every result is signal. You can open any saved item and ask it questions, or step back and see your whole library as a mind map of connected topics, the shape of your own judgment accumulating over time.
And because a personal library is personal, it lives in a folder on your Mac rather than in a vendor's cloud. AI processing runs over encrypted channels and your content is not retained on servers after the request completes. There is no account to create. The free tier holds 25 items, enough to feel the reflex form, and Mindly Pro removes the limit for €7.99 a month or €44.99 a year. The open web will keep filling with slop whether or not you do anything about it. The corner of it that is entirely yours is the part you can start building today.
Free for macOS, no account needed. Save the next ten things that prove worth keeping, and watch the one search index with a perfect signal rate start to grow. Download Mindly →