@The Old Reader: That's pretty much exactly what I'm suggesting. So one string holds the "last read on" date through to today's date and another simply counts the number of posts. The actual colour highlighting within the website (so the green bar to the left of unread feeds) will more than likely be taking place client side anyway, it just needs to then highlight everything from the "last read on" date until the current article/date. I do realise if I log in and read the latest article it might reset the counter and lose all that data, which people would be understandably annoyed about, however that's why I suggested having split date ranges or a vacation mode, so those that understand the issues with the feature can still utilise it.
Hope that makes more sense? Again, would be intrigued to know if there are some major flaws in this way of doing things?

GnorthernGnome - Can you expand on the date ranges and an unread counter concept? I'm not sure I fully understand. Are you suggesting instead of storing read/unread items per user, we should just store the date of the latest post they marked as read and then every time new items are added to that feed, if we increment a counter?

Still believe this should be a key feature and although Premium does seem worthwhile for six months I would still prefer to pay a little more to get full data retention.
I also understand the number one reason for doing this is saving server space, but can someone explain why you don't just save date ranges and an unread counter? That requires two integer strings per subscribed feed, plus two incremental counter loops. Yes, if people log in and fracture the date range it becomes trickier, but you could either have this as a work around for a "vacation mode" or pop-up a warning, or change it from x many articles to x many data ranges. Both methods = more retained data and happier customers, but maybe I'm missing something crucial with the database structure...

This is my first comment.

To freshen this discussion ... maybe ...

To the best of my understanding a feed provider has in someway defined a directory structure where its rss feed items are located. google, inoreader, and feedspot for sure and netvibes probably are accessing this. So that when I want to read further back then the current feed url displays it searches down through this directory structure. That is if I scroll to the end of the listing there is a pause and then another 40 or 50 items are displayed. Now maybe these sites are taking advantage of something like superfeedr I don't know.

Saving a search results is the defining element of a truly useful aggregator, imho. If it includes a search that goes back through the history of the individual feeds then it is truly worth money. That the search result can be exported and not require additional space on the site storage would be a reasonable compromise.

Just my thoughts.

Kim

maybe you could provide a option to download feed archive like google takeout, then you can delete it to save space in the server

At https://theoldreader.uservoice.com/knowledgebase/articles/146276-how-many-posts-per-feed-do-you-store- it says "If you don't access The Old Reader for a while..."
But it really needs to give an indication of how long "a while" is. A week? A year?

I have no desire to archive my feeds. I'm not building a library; I'm trying to get fresh updates. Please allocate resources to increasing feed update frequency instead of increasing storage of stale articles.

I just wish that when u saw these comments you'd give more storage -but not with extra charge. No, I wouldnt pay for such a thing, I werent paying for GReader, dont get encouraged by these ppl. No extras, no 1 buck for month, ok then, love you.

Yeah, That's one of the features of GReader I miss the most.

Given the storage going on now, this is pretty essential. I haven't checked The Old Reader for a few days (no more than a week) and came back to a notice saying: "You have not used The Old Reader for a while, so we kept only fresh posts for you." What??? I want ALL my posts. It's not like I disappeared for a year, or even a month. Not keeping my posts is a huge deal breaker for me. I subscribe to a lot of feeds and some are pretty busy. I often mark entire feeds read but sometimes I go through them. And I want them to actually be there.

It's one thing to mark posts as read if they're more than, say, 30 days old (what Google Reader did). But if I can't see older ("read") posts, then they're just gone, and that's not okay.

I agree that several hundred posts saved is not enough for many feeds (Huffington Post, for example). But what I find really scary is this statement: "If you don't access The Old Reader for a while, your account will be marked as inactive, and we will only keep the last 10 entries of each feed for you." What is "a while?" Considering you said I hadn't used it in "a while" when I hadn't logged in for less than a week (though I have more than 10 posts in most feeds)...

Basically you're not making me want to stay. Let alone donate to you.

I agree, I'd like to see the entire history if I want.

I'm also noticing this; when I look at my feeds in TheOldReader I only see between 5 and 20 message for some feeds whereas in Google Reader I see many many more.
But as I saw further down in this thread, maybe that is due to TheOldReader only monitoring these feeds since a few days?

Yeah, for me, that's the only problem I have with The Old Reader. When I subscribe to a feed, I also want to read the post all the way back (or at least as far as possible) and I also would gladly pay a small fee if this could be arranged.

Oh, and Pavel, you said:

"Google Reader never went and found feeds on its own so it would never index items before someone subscribed to a feed. It could only read back to the day you subscribed and that exact behaviour can be copied by The Old Reader."

That's actually not true. With numerous feeds I subscribed to, Google already had posts archived going back multiple years (and well before what was the oldest item in the current RSS feed.)

What Google did was rather intelligent. It simply archived the feed from whenever the first Reader user subscribed to it. Any subsequent person adding the same feed was only shown the posts in the current feed file when they subscribed, but if they chose to view read posts (or they searched for them), they also saw everything from that feed from all the way back to when the first Reader user added the feed.

(I'd imagine there could potentially have been holes in that coverage, if there were periods where no Reader users were subscribing to the feed, but it was still much better than nothing.)

There's no reason The Old Reader couldn't replicate that same behavior.

I, too, would happily pay a few bucks a month for this feature.

Bigger storage space for The Old Reader, and the ability to search back in a feed as far as the original feed provider allows, would be great indeed. And yeah a tiny fee to cover server and maintenance costs would seem reasonable if it can keep adverts away.

Would pay for this too!

That's the feature I can't live without. Many of my feeds are from closed site.

And I believe it is at least technically possible.

Details:
Cached items of google reader can be accessed without auth, like
> http://www.google.com/reader/public/atom/feed/http://www.digg.com/rss/index.xml
And a few non-offical GR api uses this as well.

I myself tried download all the xml cross-site-ly by
- a reverse proxy elsewhere
- or browser extension ( the chrome version : http://jokester.github.io/feedzombie/ )

BUT in either way, google would start to ask for recaptcha, after hundreds of requests.

Given a central server to submit to, maybe we get them all in a crowd-sourcing way.

I'll pay for this too. Right after "show only feeds with unread"