Reeder deleted the unread count. I counted what it cost
The update lands. The unread badge is gone. You open Settings, then Appearance, then every menu that looks remotely plausible. There is no switch. The app did not misplace your unread count. It decided you no longer need one.
A 2-star Reeder review gave me the sentence I could not shake:
But letting go of unread counts feels like letting go of the edge of the pool in swimming class for me — I don’t want to.
That is why I spent a weekend pulling 1,811 App Store reviews for eight feed readers. I wanted to know whether the anger was ordinary resistance to a redesign or whether Reeder had removed something people were using on purpose.
I should put my bias on the table. I am building a feed-reader sync service, so a finding that says existing readers are leaving people behind is convenient for me. Read the numbers, not me.
The number was never just a number
“Give me the unread count back” sounds like a feature request. It is smaller than the thing people actually lost.
An unread marker is a piece of memory. It tells you that you have already considered an item. The count tells you where you are in a queue you may work through over several sessions. You can leave one folder, check another, come back tomorrow, and still know what happened.
When that state disappears, the app has chosen a reading method for you. Start at the newest item, scroll down, and trust the saved position. That works if you read one continuous timeline in order. If you move between folders, skim ahead, or return to an older story, the position cannot answer the basic question: have I seen this?
The obvious workarounds all change the job. Saving every interesting item forces a decision while you are scanning. Treating the scroll position as read state assumes one path through the feed. Staying on Reeder Classic is a perfectly sensible answer today, but it does not make the new model fit the old workflow.
This is the difference between the two models. An inbox gives every article a read or unread state and counts what remains. A river of news is a timeline you dip into and leave without an obligation to clear it.
Reeder solved a real problem
The river model is not a design mistake. For some people, the unread count had become a machine for manufacturing guilt.
A 5-star Reeder reviewer described the old model this way:
I follow a lot of feeds, and while the “old” Reeder was great at helping me keep up with what I cared about, it created another inbox for me to clear.
Remove the count and there is nothing to fail at. You read what is in front of you, save what deserves more time, and leave. For readers who had started avoiding RSS because the backlog felt like debt, the timeline is relief.
Reeder is also an excellent product. The new model is deliberate, coherent, and defended by people who use it. Calling those people casual readers or pretending they have not understood RSS would be an easy way to avoid the actual disagreement.
The trouble is that anecdotes cancel out. One person feels liberated; another feels lost. Reviews of a redesign are especially messy because people are reacting to navigation, density, folders, sync, and price at the same time. I needed a way to see whether read state was a recurring fault line rather than the loudest complaint in a handful of memorable reviews.
My first count was wrong
I scraped roughly the most recent 500 US App Store reviews available for each of eight readers, collected on August 22, 2026. Some apps had fewer. The full corpus contains 1,811 reviews.
Then I searched for references to unread counts, read and unread markers, marking items read, and the absence of those controls. My first pass also caught “read it later.” That inflated the result by about a fifth. I removed those false matches and classified the remaining reviews by tone.
One limit matters. Recent reviews skew negative because people tend to write when something breaks or changes. These numbers show direction and support comparisons between apps; they do not tell us what percentage of all users is unhappy.
Inside Reeder, 61 of 312 recent reviews mention read state: 19.6%, or about one in five. Of those 61, 28 oppose the change, 23 support it, and 10 are neutral. Among the reviews that take a side, the split is 55% against and 45% in favor.
Neither side is small.
The critics were not all asking for a red badge. A 2-star reviewer made the distinction plainly:
I don’t care about an unread count but I do want to see what I’ve read or not, that’s much superior to scroll position.
Another 2-star review asked for the compromise the argument keeps missing:
I like a lot of what’s new and improved in this app, but having 100+ articles and no read states is impossible to deal with. Please give us the OPTION to toggle on read states.
That uppercase “OPTION” is the whole article in one word.
The same choice produces the same complaint
Reeder's 55/45 split shows a divided audience. The comparison across apps is harder to explain away.
The table below reports the share of each app's recent negative reviews that mention read state. It does not report the share of users, or even the share of all reviews.
| App | Model | Negative reviews mentioning read state |
|---|---|---|
| Current | river | 24% |
| Reeder | river | 22% |
| Unread | mixed | 19% |
| Inoreader | inbox | 6% |
| NetNewsWire | inbox | 3% |
| Feedly | inbox | 2% |
| Readwise Reader | inbox | 0% |
The two river products land at 24% and 22%. The four inbox products range from 0% to 6%. Unread, which mixes the models, sits between them at 19%.
Current and Reeder are independent apps with different interfaces and users. They made the same structural choice and received almost the same concentration of complaints about it. The inbox readers did not. Reviews are not an experiment, but this pattern is not a personality clash between Reeder and its longtime customers.
It is the cost of making one reading preference mandatory.
Let the reader decide
There is no need to vote inbox or river off the island. A reader can preserve read state underneath and decide how much of it to show.
Inbox mode can expose markers, counts, and commands for clearing a queue. River mode can hide all of that, remember the scroll position, and let old items drift away. A default can still have an opinion. It just does not need to trap the user inside it.
Lantery is in development and is not publicly available. I am building it with inbox and river modes as a user setting, plus sync for existing feed-reader apps and a web reader for computers where those apps do not exist. That makes this finding useful to me; it does not change the count. The current project status is at lantery.app.
The river model is not wrong. Making it mandatory is. About half the people who spoke up were using the old model on purpose.