December 16, 2004

Don't F*** with My Data

UPDATE 12/17/04: I posted below a rant on the loss of data, that was too harsh. I admited in comments below.. but I wanted to state it at the top. Many commenters feel I should have had more than the one backup of data, and others feel the software failed. However, I was angry and shouldn't have posted in that state, but I was annoyed due to the second loss of data that a backup copy didn't solve. Please read the follow up post, asking about what and how users and developers are responsible to each other for the purchase or sale of software.


The person sitting next to me just said (she's requesting anonymity), "... usually when you're paying to get fucked, you get it how you want it." Well, NetNewsWire has totally messed with my data. Twice now.

The first time, I told them directly in email. Their system is set up to do a 30 day trial.. and who remembers exactly when the 30 days are up, right? But instead of alerting me with a notice that the 30 days were about to be up, before the system expired, they simply told me after the fact. No problem. Pay for license. Like it better than Shrook and two others I was testing. Big deal. EXCEPT. When they synced my local copy of NetNewsWire with their server, to make sure I'd paid for the license, they lost all my data from the previous 30 days.

I emailed them. And they knew they blew it, offering a refund, and said they'll fix this eventually. But they don't understand! I'm tracking data on tons of services, people, companies, websites, blogs, projects, as well as reading feeds. This is fucking social media after all, people. And it's my work, professional and academic. It's a huge part of what I do. This is aweful!

I don't want a fucking refund. I NEED the data. Moronic.

Anyway, now, yesterday, I realized that NetNewsWire had deleted all the entries on few of the meta search feeds, for the past two and a half months, retaining only the most recent searches. I have the system set to save all data for 999 days. ACK!!!!! It appeared only on Technorati watchlists at first, so I IM'd Dave Sifry last night to see if it was them. Can't tell.. but since this happened in the last two days (I happen to have screen shots of NetNewsWire from two days ago showing that all the posts from the last 2.5 months were there as of then)... I wasn't sure if they changed something or not. They are checking, but now, it's appeared with some other feeds like my Engadget feed, which only goes back 24 hours. I should have 5 months of Engadget posts saved in NetNewsWire. I want that kind of data to search and compare old posts to current ones. So I'm not sure if NetNewsWire changed their GUID structure or these feeds did.. but considering the feed data, including old data from my summer use, plus the last couple of days, are showing up correctly in Shrook, I'm thinking it's NetNewsWire's problem.

ANSWER: I'm paying for the Shrook license now, because I need a backup, because I can't trust NetNewsWire. But I will have Shrook's data from two months this summer, added to current data, which a huge 2.5 month hole in the middle.


Anyone know how to export the data (not just OPML feed list) so that I can save it somewhere for safe keeping?

Posted by Mary Hodder at December 16, 2004 02:28 PM | TrackBack

Memo to self: Never, ever, under any circumstances, penalty of death, or even triple dog dares, piss Mary off. :)

In my limited experience, Brent Simmmons, NetNewsWire's author, is a good guy. If your data can be retrieved, I'm sure he'll do whatever it takes to retrieve it.

If you want to find old posts from public feeds, you can look them up in Bloglines. We archive stuff from the time we start crawling them, although currently you have to be subscribed to them in Bloglines in order to view more than the most recent 20 articles. If you send me a list of the feeds you need, I can quickly do a dump of them in RSS format.

Posted by: Mark Fletcher at December 16, 2004 04:20 PM

What version of NetNewsWire were you using?

Posted by: Jim Roepcke at December 16, 2004 04:52 PM

Mary, I understand your frustration, but I've been on both sides of this issue a few times, and the reality is, NetNewsWire is the product (basically) of a husband and wife couple.

They're good people. They don't want to screw up, and I assure you (without even having spoken to them, and only ever having met you once) that they didn't want your data to be lost. And they made an app that you thought was good enough to pay for, so they're clearly not morons. Most likely, Brent's human, and he made a human error, and while it's perfectly reasonable to not want to pay him for software that doesn't work like it should, I'd also hesitate to be *too* abusive towards people who made an honest mistake.

Most of all, the overwhelming majority of the content you've collected is information that was posted publicly to the web. I'm certain, especially given the large number of people trying to build businesses around syndication feeds, that you should be able to recover the data you've lost. Do any of the Technoratis or Feedsters of the world let you pay for that? They should.

Posted by: Anil Dash at December 16, 2004 05:34 PM

Hi Mark,
If you make Bloglines available to me locally on my machine.. for when I'm traveling.. I'll use it. Do now use Bloglines for a bunch of stuff... but not as my main aggregator. Despite my desire for 24 hour-a-day wifi worldwide, it unfortunately doesn't exist yet. So I need to be able to take it with me, search it, etc.

I'm sure Brent Simmons is a lovely person, but I paid for his stuff and it's not working. It's not a matter of a refund. I'm depending on his service to get it right because it's bringing me data about and for everything I do. Sorry to get angry, but it is really awful to depend on something and have it fail repeatedly in different ways.

Oh, and Jim, my version of NetNewsWire is 2.0b3.

Anil, the thing is... by charging, they imply that they are professionals. If the service isn't reliable, then they should keep it free, until it is reliable, at least as far as holding onto the data the service collects. We buy it to collect data, afterall. And the issue with back data is, once those posts fall off the front of a blog or other RSS content maker, it's harder to get in order to put it into little buckets to search or analyze locally. Because the information is only located on other search services at that point, they aren't searchable a second or third time across increasingly small buckets. I'm not trying to be abusive. I just believe that if you ask people to pay for something, you should be prepared to deliver. And finding out that a ton of data disappeared, leaving a hole between Sept 20 and yesterday, across approximiately 50 search feeds, is very upsetting.


Posted by: mary hodder at December 16, 2004 05:37 PM

Sorry, that was "only ever having met *them* once."

Posted by: Anil at December 16, 2004 05:38 PM

I've used NNW since it came out and haven't had any trouble, but it sounds like your really a power user. If it's that big a part of your life you really need to have proper backups in place and not depend on the app. I don't depend on Word not to mess-up a file or my hard drive not to crash when I'm working on a program. Unfortunately the technology just isn't there yet to do most of what we do without backups.

In any event I think the data you need to backup is here (at least in v2):
/Users/yourmacusername/Library/Application Support/NetNewsWire

As a side note there is alot of other apps data in the Applicaiton Support folder you'll probably want to backup while your in there. That's a common place for apps to store data.

Posted by: Ian Landsman at December 16, 2004 05:42 PM

This reminds me of the, perhaps apocryphal, story I heard a few years back about a disk crash on a Stanford server used by many graduate students. Some of the users were storing literally years of work on the machine, and, in a frightening number of cases, it was the only copy in existence. Fortunately, pretty much all of the data was recovered. As far as I know, no suicides or permanent nervous breakdowns occurred in the two days between the crash and the recovery. To this day I still wonder how someone could spend months or years working on a project, and never once think to make a backup.

With this in mind, I ask, "Mary, what the fuck were you thinking?"

Really, I don't mean to add insult to injury. You're right, losing data sucks, and it sounds like NetNewsWire really did fuck up. If you'll forgive my (well intentioned) chastising, I'll gladly make amends by, if possible, helping to recover what's lost, and cooking up a nifty automated backup scheme to make the next time something like this happens suck less.


P.S. You might try poking around in:
~/Library/Application Support/NetNewsWire
The FeedSource and FeedData directories, as well as the History.db file look promising.

Posted by: Daniel C. Silverstein at December 16, 2004 06:09 PM

Hi Ian,
So, actually, backing up, in this case, didn't help (though I was backing up because of the previous data loss). As I restored the old file, which is in a proprietary format to NetNewsWire, it immediately synced on the messed up 50 feeds, and lost all the older data again. And somehow the backup file got corrupted and now I can't do anything with that. So having a whole other service, with main and backed up files for that, seems like the best way to do it. That way, I'll have two services locally and two backup files.

I've also exported my OPML file to bloglines for web service on these feeds.

Posted by: mary hodder at December 16, 2004 06:11 PM

Thanks for answering my question Mary. The "b3" in 2.0b3 stands for "beta", and "beta" is a specifier in software versions that means "not ready for production use", or "still being worked on, there ARE showstopper bugs in this software that prevents us from calling it '2.0'".

A recommendation for going forward: if this data is as mission critical as you're describing (and I believe it is), I strongly suggest you set up some forms of redundancy, and a backup strategy, to avoid these problems. Also, don't use beta software for managing important data.

For redundancy, you could use multiple clients. For example, NNW and Shrook. You seem to like the NNW interface better, so use that, but keep Shrook running too. If NNW goes down, you have the data in Shrook and your work doesn't get affected. Sounds like Bloglines or other similar web-based services would also be good ways to create redundancy.

For backups, find out which files and folders NNW or whatever client you use going forward writes to, and configure some (production quality) backup software to back up those files and folders regularly. It's critically important that you test restoring the data and test the restored data. There's nothing more frustrating than thinking you have a backup when you in fact restoring doesn't work or restoring results in something that still doesn't work.

I too know Brent, I've known him online since '96 and finally met him in person last year. Brent is one of the most, if not THE most honourable and ethical software developers I've ever met. That is not hyperbole. That Brent can be an all-around good guy AND apparently make a living doing it makes me respect him all the more.

I think calling Brent names because of data loss caused by misusing beta software is unfair. Look at the top of the page you downloaded NNW 2.0b3 from:


Beta software has bugs! Nasty, vicious bugs with great big, sharp teeth!

Don't use beta software unless you're clear on what "beta" means and you’re comfortable running beta software."

You _were_ warned. I've lost data too, it sucks. I wish it didn't happen to you, but it's not Brent's fault.

I hope you follow my advice and take steps to prevent future data loss problems.

Posted by: Jim Roepcke at December 16, 2004 06:16 PM

Hi Jim,
Okay, I'm changing the post. I got really mad when I saw that it wasn't just the meta search feeds that were messed up, but lots of others. You are right. I shouldn't call them names. In fact, what happened is moronic. Done.

About the beta aspects.. why charge for beta if it's not ready to actually hold onto the data that the service is supposed to be collecting? I feel that by charging, they are cueing us to think the stuff is ready. Beta/NC means to me that I should take extra steps to have redundancy beyond backing up NNW files. NNW for money means they are ready for us to take them seriously.

Also, I didn't download this from their site. It was sent to me to try. I didn't opt for the beta as you suggest. I realize that beta means it can fail. But it wasn't one feature that failed. The system seems unable to retain data on some feeds. That's a complete and utter failure of the main functionality. So, why do they charge for beta, then, if it doesn't yet work?


Posted by: mary hodder at December 16, 2004 06:37 PM

Not meaning to be totally unhelpful, but it sounds like you are using RSS reading in a way that is beyond the general "RSS usecase" of: the most current news, delivered to your desktop, ASAP.

I have needs for building up information from websites into historical and/or categorical stores, and I don't see the RSS / newsreader combo (being so current-time oriented) as being designed for my needs. It can be helpful, but I try to pull stuff out of RSS / news readers ASAP (into a database is handy, if still pretty inconvenient).

It sounds like you need something outside the typical RSS reader box, so you'll need to be really extra careful with anything that is designed to be a RSS reader, at this point at least.

Posted by: Jay Fienberg at December 16, 2004 06:43 PM

All of this over a $25 piece of software. What a fucking asshat.

Posted by: Dave at December 16, 2004 06:43 PM

I have to agree with Jay, this isn't the way most people expect a news aggregator to work. I agree with you that it SHOULD work, but it's probably not a very well tested usecase for it.

Posted by: Ian Landsman at December 16, 2004 06:47 PM

I can only speculate why they charge for the beta. Usually companies don't sell betas, they sell production copies and either offer betas to paying customers or offer free open betas. The 2.0 beta is available as a free demo. Before now, I just assumed it was a 30 day demo because they didn't want you using beta software longer than that, and they intended to release new betas (or the final version) within 30 days of any prior beta release.

I would only guess they let you buy it so that you don't have to buy a license for 1.0 and then bother with upgrading it. Dunno. It is cheaper than NNW 1.0, so I think it's a good deal.

It's still called beta, and that has very serious connotations. It's unfortunate your channel for the software didn't warn you about its suitability for production use. But you did opt for the beta, because you ran it and relied on it instead of a non-beta version. Anyway that's entirely moot.

Beta software doesn't have to only have one bug or one feature that fails, it can have dozens, or nothing can work! Note that 2.0b6 is out now, it might have fixes for the problems you had, but it might not. I'm still running 1.0.8 because I'm too lazy to deal with the ramifications of having NNW fail on me. 1.0.8 is rock solid for me, but might not have all the features you need for your work.

Best of luck in the future!


Posted by: Jim Roepcke at December 16, 2004 06:49 PM

The unfortunate thing is that the term 'beta' is used far too often (just look at every service Google offers besides search), and has lost its meaning.

But even though it was beta software, I don't think blaming the victim is very productive (or correct). There are many things software developers can do to protect against data loss, whether that loss is due to bugs or otherwise. I think it's a sad commentary on the state of software these days that there isn't more outrage over this type of misbehavior in applications.

Posted by: Mark Fletcher at December 16, 2004 08:38 PM

If your clothes got dirty because you sat in a restaurant that was under construction, would you blame the restaurant owners and construction workers, or think, "maybe I should have waited for the restaurant to open before wearing my favourite suit there"?

Beta software is under construction, and that means its dirty and not ready for people who can't handle cleaning up messes.

A unique and useful thing about software development is, at some point it becomes possible to let interested people try the software. From the developer's point of view, those people will hopefully provide feedback and/or bug reports.

There is risk involved, just as there would be using any product that wasn't ready for the general public, be it a car, a pair of skates, hair colouring formula, medicine, etc.

Brent's beta software download page had plenty of warning about the software's correctness. More stable production quality software was just as easy (if not easier) to find and download.

That the term "beta" has lost its meaning is not the fault of developers, its the fault of software users who have lowered their expectations from software. Beta software usually doesn't break badly, so people may be lulled into believing beta means "good enough". It's not, it just means those developers have pretty high standards and you've been lucky so far.

If it says beta, and you're going to get upset if it breaks, PLEASE stay away from it, it's _not meant for you_.

If Google says their stuff is beta, it's because it's unfinished and may have more bugs than their usual production-quality software. They offer it because they want feedback. That it is actually useful in a beta state is a testament to the quality of their software, not a statement on the value of the term 'beta'.

As a software developer, I understand better than most people how badly incorrect software can damage my computer. This is why I choose to avoid beta software as much as I possibly can on my main computer. If my main computer is not working, I'm in a world of hurt. If it's not working because I didn't respect the fragility of the system, ultimately it's my fault. It's always easier to blame others, but that doesn't make it fair or right.

I do think it's a failure of the software industry that users aren't better educated about computer maintenance, and that's something I hope will improve over time. I don't have any suggestions off the top of my head for improving that situation. That said, I think Brent did all he reasonably could to discourage people from relying on his beta software for critical tasks.

Posted by: Jim Roepcke at December 16, 2004 10:16 PM

Another comment about 'beta' in the context of Google and other "platform developers". Beta also means that the developers has the right to radically alter the functionality of the software, and its public programming interfaces, before the software is declared "final".

This is critical, because generally when you ship something as final, and it is scriptable or otherwise programmable from other software, the developer should try as hard as possible to maintain backward compatibility in future versions. Therefore, if the design of your software stinks during the beta, you still have the opportunity to get it right before you commit to supporting that functionality down the road.

In Google's case, they let developers access their back-end data through web services. If they say a service is beta it means, "don't blame us if we get a better idea, change this thing, and it breaks your software that connects to ours".

Once it's no longer beta, customers have a legitimate complaint if changes break their stuff.

If you're not a developer or a system integrator, it's easy to overlook that and not understand why it's important to call some things beta when they really are very solid and for all intents, ready for public use.

Posted by: Jim Roepcke at December 16, 2004 10:25 PM

Rule one of Mission Critical software is "once its working NEVER EVER EVER upgrade." Now you know Rule One. Welcome to the club. You may also want to look into your backup regimen since it clearly doesn't work---"screwing up" a backup file shouldn't ever happen. Yeah, it sucks but there's nothing else the Simmons' can do to help you if your data is trashed.

Posted by: Byron at December 16, 2004 10:35 PM

Thanks for the comments, Byron and Jim. I think you're missing the forest for the trees.

The point is, don't charge for software, knowing you have bugs that will lose data. Even if you are in 'beta' -- charging makes it seem like it will work reasonably well, and as an 'aggregator' you ought to be able to hold onto the data. Not charging would cue us to think it was an actual 'beta' meaning 'there are still major bugs' -- even bugs that as 'an aggregator, mean we will lose data regularly.'

I don't think I need to apologize to NNW for calling them on this problem as Jim suggests in his blog post, or saying that their software is not trustworthy. Yes, I went to far in referring to the makers of this software as 'morons' which I changed, if you read earlier comments, to 'moronic,' which is what I meant to begin with.

It is absolutely stupid to sell software and then hide behind beta status to justify knowing you have bugs that will lose the data out of your feed aggregator. Google doesn't sell Google News -- it's in beta. Fine. It's free. NNW is not free. They seriously need to rethink their payment plans as it stands with software quality.

Posted by: mary hodder at December 16, 2004 10:58 PM

I agree with Mark, 'beta' is way overused. In the case of NNW it is entirely appropriate, but others have muddied the waters.

Posted by: Ryan King at December 16, 2004 11:21 PM

Mary, I thought you should apologize for calling the people at Ranchero morons for something that ultimately shouldn't have happened, not for calling them on the problem (though there are better channels for reporting problems with beta software). I agree the situation is moronic. A comedy (tragedy?) of errors, with errors on both sides.

Where we still disagree is on expectations. You expect the software should work because you paid for it, I expect the software won't because it's called beta. Regardless, Caveat emptor!

All software has bugs, period. You said they hide behind beta status to justify knowingly releasing software that loses data. Beta status, by definition, justifies any software misbehaviour! It's unfortunate you weren't informed ahead of time.

I do agree it's unorthodox to sell a license for beta software and from that angle you have a legitimate complaint (although they did offer you a refund, there's nothing they can do if your backup became corrupted(?!)). The beta you were using was out of date, and it's unlikely the problem you saw afflicts most beta testers or the beta release would have been pulled (or updated promptly, which is possible, but you unfortunately didn't update).

It would be helpful if their online store explained the beta status like their beta download page does.

Hopefully this thread has helped you understand beta software better, in general, and ways to avoid similar problems in the future. Software developers NEED users like you that understand how to extract value out of new ideas and feel strongly enough to pay for software, but obviously we also need to do a better job of managing expectations.

Posted by: Jim Roepcke at December 16, 2004 11:36 PM

Ryan, change 'beta' to 'beta software', and you've hit the nail on the head.

Posted by: Jim Roepcke at December 16, 2004 11:40 PM

All software has bugs.

Every application from every developer, from Microsoft down to a single coder working part-time out of his home, has bugs. Not just beta versions, but also 'final' releases.

Countless thousands of people have lost/corrupt data at the hands of non-beta versions of high profile apps like Excel, Word, or the underlying OS - all of which we pay a lot of money for... much more than $25.

Even mission critical software like oracle has corrupted data because of bugs within oracle itself. You're potentially talking about over $1,000,000 worth of database software (oracle 9i, 32-cpu license) - and it comes with bugs.

It's great that you had a backup, but you didn't follow the first rule of backups - get them off the same system as the original data. If it's small, archive it to removable media, like a CD. When you need the backup, work on a copy and keep the original backup safe.

I don't see how you can trust any software, if you're expecting a perfect product - it just doesn't exist.


Posted by: Mark at December 17, 2004 12:13 AM

"The customer is always right."

Posted by: MT at December 17, 2004 12:28 AM

This talk of all software having bugs... as Homer Simpson might say, "That's Losertalk". Is it acceptable for a Civil Engineer to say that a building will fall down at any given moment, when you least expect it? Is a Mechanical Engineer exempt if the jet engine they signed off on showers the passengers of the plane in fan blade fragments?

If you've answered "no" to either of those... what makes computers and software developers special? Computing technology has become an integral part of our lives in a very short time - why should practitioners not be held to the same standards as other, more established professions?

Posted by: Randy at December 17, 2004 01:03 AM


A couple of times you've mentioned not charging for beta software - but in some cases where the safety of the data isn't really important beta software is usable by those of us who want the new features.

It is certainly true that companies like Google and Microsoft are diluting the danger implicit in the use of the word beta.

In future I'd be more careful of using software that is marked as beta (no matter who sends it to you), beta software is dangerous and not just to the data held in that app.

Posted by: Ross at December 17, 2004 02:37 AM

How do you know Brent *knew* there would be bugs that would result in data loss? Perhaps he just knew that it was still unfinished and that it would probably have bugs in it, and wanted some help shaking them out?

Also, would you hold a civil engineer responsible for earthquake damage to a bridge in an area that isn't near any known faults and has no known history of earthquakes?

Complex systems cannot be designed for every possible contingency, there's this thing called "combinatorial explosion" that gets in the way. It's the same reason you can't play perfect chess by simply computing every possible game either in advance or on the fly.

What people can do is make a good-faith effort to take reasonable steps to create structures that hold up well under common conditions. Edge cases are always a possibility, and rigorous analysis *coupled with* rigorous testing can help minimize them.

Posted by: at December 17, 2004 03:07 AM

Geez you have a potty mouth! Do you eat with that thing!?

Posted by: at December 17, 2004 05:31 AM

I know that NNW knew about the bug leading to the first data loss (after the 30-day trial, when the system syncs with a newly purchased license, all previous data is wiped out). This is because they told me it was a known bug, after I told them about the data loss and asked for help getting the data back. They asked for my files, which I sent. They didn't respond with an answer, until my third request to find out what was going on, wanting to know when I'd be getting the data back. At that point, they said that the data was in fact wiped from the files and they couldn't help.

So my question in my reply email to them was, if you know you have a known bug, that occurs as you are syncing my copy of NNW with the mother ship, and sending me a popup notice to buy the license, why not include some text below the license info like, "Caution, when you buy the license, your data collected during the 30 day trial will be lost. Please follow these simple instructions to prevent this: 1, 2, and 3...."

It's cluetrain. This is isn't Microsoft. Social media means you treat me like your partner, not someone you withhold information from, because you are afraid of telling the truth ... let me know how to prevent data loss, and I will have no complaints.

I'm pretty sure by now NNW knows about this post. If they don't, then they really do have problems greater than dataloss and they aren't long for this world. I have to give it to Dave Sifry, who shows up and chats directly in email or on blogs when Technorati has a problem for a user, and that service is free. Take some responsibility.

If you sell an aggregator, chances are I'll buy it to aggregate, or collect, data. Know you have bugs that lose data, but don't inform me, and I'll be mad, warn others about it by blogging the whole thing on my blog, and never buy your stuff again.

Posted by: mary hodder at December 17, 2004 06:41 AM

I've been using NetNewsWire since the beginning and I never lost any data with it. I'm now using the latest beta version, and with one beta there was a notice that the data format changed and you won't be able to go back to an earlier version.

If your data is so critical there's also a stable 1.0 version you should have used instead of the cutting edge beta.

Posted by: Mike at December 17, 2004 06:46 AM

"Also, I didn't download this from their site. It was sent to me to try. I didn't opt for the beta as you suggest"

Posted by: at December 17, 2004 07:22 AM

Seeing as the Beta Software was for Licensed Users Only, you really don't have a leg to stand on here. The Beta run was for people who already had licenses, any data you lost because of this whole thing is pretty much your fault since A) you weren't a licensed user for the beta and B) it's beta, which means it's volatile and non rock-proof.

I'm sorry you lost your data, but really, you can't entirely blame Brent for it.

Posted by: Tom Bridge at December 17, 2004 07:23 AM

To ride the cluetrain you have to have a clue. (If) You're not a licensed user of software that you didn't pay for, no one is going to take you seriously when you talk out of the other side of your mouth how important it is to you. Either you're trying out new 2.03xxBb software, or you're using 1.0 production software.

I know Brent by reputation: in his place I wouldn't lift a finger to help you, but I'd bet he does. In the meantime you just keep on wailing. The cluetrain tracks travel in both directions.

Posted by: George Girton at December 17, 2004 08:30 AM

"Also, I didn't download this from their site. It was sent to me to try. I didn't opt for the beta as you suggest."

Thank goodness they didn't send you a really cool screensaver.

Anyway, the Ranchero Site is abundantly clear about the potential dangers of beta software on its download page for NNW Betas:


Beta software has bugs! Nasty, vicious bugs with great big, sharp teeth!

Don’t use beta software unless you’re clear on what “beta” means and you’re comfortable running beta software."

Don't blame Ranchero for the fact that you chose to run software that seomeone sent you to try without bothering to investigate exactly what you were running.

Posted by: Brian Carnell at December 17, 2004 08:35 AM

You people are insane. Yes, a software developer can put out betas, with explicit warnings and all. But that doesn't mean the developer can't spend the extra time needed to add in automatic backups and other features as a form of defensive programming to proactively guard against these (and other unexpected) types of problems, especially in beta software. I know some other aggregators do just this. Do they have to do this? No. But should they? Absolutely.

Put another way, as a software developer, do you want word to get out that your application loses data? It doesn't matter if it was beta software or not, because that distinction is lost on the 'average user' (not the people in this thread). If you get such a reputation, you're dead. Hopefully that isn't the case here, because Brent is a good guy and NNW is a good product. But wouldn't you want to do whatever (extra) work was necessary to try to prevent such a situation in the first place, even in beta software?

So, in this case, is Brent wrong? No, not really. But is Mary wrong? I would say definitely not.

Posted by: Mark Fletcher at December 17, 2004 08:41 AM

No wonder software sucks so badly these days. Try to do something useful by pointing out some difficulties you are having with it and you get attacked.

The user is almost always wrong, but the user pays the bills. Looks like Mary has paid a huge bill here and deserves better.

So, if what we're doing here is playing around and having fun like monkeys, then great. If we are trying to make software better, then let's get serious and get to it.

P.S. Bloglines rocks my world!

Posted by: Tom Ligda at December 17, 2004 08:43 AM

You were lucky in comparison to me. NNW crashed and hosed my Safari browser bookmarks file. I lost weeks of work. I had to go through the browser cache manually with a text editor and retrieve what little data I could, it took days, and I still lost most of my bookmarks.

Posted by: Anon at December 17, 2004 08:55 AM

Hi George. I am a licensed and paid user of 2.0b3 NNW. I'm not 'trying out' that software. I bought it. If NNW doesn't want to be responsible for the software, then don't sell it. That's the point of this post.

Hi Brian,
I was sent the software to try, and then purchased it. Did not see, though it may have been buried somewhere in the purchasing process, that bugs are nasty with teeth. My point is not that the software is beta or has bugs... my point is that bugs they know about before they make the sale should be told to the buyer.

I don't consider it wailing to warn others of these issues since NNW isn't warning people specifically on their site. They may say there are possibly nasty bugs, but losing all the data is a problem central to the point of their service. They know how to prevent it, have the opportunity during the license popup process, and yet don't do it. When I suggested that in email they blew it off. I'm not sure they really care about letting people know how to deal with the problems. My next post will constructively do that.


Posted by: mary hodder at December 17, 2004 09:01 AM

I can see why you're pissed, but I don't think you're doing a very good job of seeing this issue from the developer's perspective.

If you didn't see the "Nasty, vicious bugs" warning, how could Ranchero have communicated the issue with the known bug to you?

Every good software developer knows about more bugs than they have fixed, especially if they are keeping a bug database.

While I like developers to make this database available to the public, most if not all general consumers would run screaming from that information if it were presented to them. No one cares about a bug until after it bites them on the ass. Though you can see in hindsight that this particular bug is of such earth-shattering risk it should have been given a pop-up dialog at installation, there are probably others that caused other NNW users to rain down a little Old Testament wrath.

Any time Ranchero spent dissing its own software with bug warning dialogs could have been used to fix those bugs. Which would you rather have him do?

Posted by: Rogers Cadenhead at December 17, 2004 09:19 AM


You've dished it out in your post and taken your fair share of abuse, too. If there's a flaw to point out in the "system" here was the person who sent you the beta software. If you had been sent to the developer's webpage to download the software yourself, you might have had a different experience.

As someone who works for a software company, this whole experience reminds me that our users are more precious than we think. It also makes me reconsider public beta testing.

Posted by: Steve Kirks at December 17, 2004 09:38 AM

In response to Tom Ligda's comment above, I don't think anyone is being attacked and it is certainly *NOT* Developers vs Users. You may well be in the 'developers are evil' camp but that is a generalisation.

Any comments you might construe as being 'aggresive' are in fact defensive so that somebody's good reputation that is well deserved is not ruined by a single person who was using software unwisely (by using a Beta). Sure, maybe he shouldn't have sold it but as far as I can tell Ranchero is a very small company - you put up with worse from Microsoft, and everyone is entitled to one mistake.

There is more than one person at fault here Mary, your friend for suggesting you use Beta software (when they possibly knew how important it was you didn't loss data), you for not keeping a backup (if you had you could have asked Ranchero to give you another 30 day trial and work from your backup), and Ranchero for selling you a licence for Beta software.

No doubt there will be excuses that are valid for all of you, your friend didn't know what beta software means, you for not knowing how or what to backup and Ranchero for assuming you knew you were buying a beta licence.

On the positive side, you've learnt something.
On the negative side, you've damaged a good company's reputation. There are bigger companies more worthy of your ire.

Posted by: Ross at December 17, 2004 09:40 AM


Your experience with the software demonstrates that the product is not ready for prime time. You should stick with more experienced vendors.

You should never allow proprietory formats on your machine, from any vendor. This is why YOUR backup of YOUR data failed. Let that be a lesson.

For Ross: the company got what they deserved. They released a buggy product to the public and got bad review and "constructive" consumer feedback. Just like Oliver Stone's Alexander.

If I was the president of NetNewsWire, here's what I would do: I would send her a check for $5000 for making up for her loss of productivity (she's going to have to redo most of that work). I would also send her a personal, handwritten note telling her like this:

"We have learned a lot from your problems with our software, and do deeply apologize for any inconvenience this has caused you.

I personally want to assure you that we will make every effort to never let this happen to any of our clients, yourself included.

I have included a check for $5,000 to you for compensation for your efforts in re-creating your data and teaching us this lesson. "

Of course, it would be unrealistic for this to have happened back then. Now, in hindsight, it's obvious it should have been done.

Posted by: Christopher Mahan at December 17, 2004 10:09 AM

Christopher, of COURSE it was buggy. Its a BETA. I don't see how trying your best to ruin someone's reputation is considered constructive anything.

I'm not saying Ranchero are blameless but the fact they made a mistake does not excuse the user from the their of the bargain ... "I shall not play fast and loose with my data, if it is important I shall make the effort to back it up".

Posted by: Ross at December 17, 2004 10:18 AM


As Mary so poignantly pointed out, she was charged good money for it. It is absolutely normal to assume that once you pay for something, it is going to _work_. The company should not have charged her for the software.

Are we living in some sort of alternate reality today? You buy a car at the dealer for $35,000, and it blows up in a fireball killing your family, and the dealer says, "oh, sorry ol' chap, it was beta. You shouldn't play fast and loose with your family like this."

WTF? If it's beta, dude, you DO NOT charge for it.

The company made the decision to charge for their beta software, not warn users about possible data loss, and now they're reaping the consequences. I'm with Mary on this one: let them suffer.

As far as backup: She did back it up! She tried to recover the data, and the softtware blew that too! She can't get it any other way, since it's in a PROPRIETARY format. What, I play tell, should she have done differently?

Posted by: Christopher Mahan at December 17, 2004 10:34 AM

Christopher I am not attempting to justify charging a user for a beta - I think its wrong and I believe that public betas are almost always a mistake. However, I wouldn't call doing a backup in *response to data loss* a very effective mechanism, but we live and learn - I am sure Mary will not repeat her half of the mistake again :)

{quote}backing up, in this case, didn't help (though I was backing up because of the previous data loss). ... somehow the backup file got corrupted and now I can't do anything with that.{/quote}

Posted by: Ross at December 17, 2004 10:51 AM


There are no current known data loss bugs. There was a bug in 2.0b3 that happened when the demo expired, but it has been fixed.

(Also, to be clear -- NetNewsWire does not ever "sync with the mother ship.")

But just because there are no know bugs doesn't mean there aren't any. I'd like to retrieve your data and figure out and fix the bug -- but I need your help to do so. If you'd answer a email me a few things, that would help tremendously:

Question: which subscriptions lost old entries?

Things to email me:

1. Prefs: ~/Library/Preferences/com.ranchero.NetNewsWire.plist

2. Subscriptions: ~/Library/Application Support/Subscriptions.plist

3. Data folder: ~/Library/Application Support/FeedData/

Posted by: Brent Simmons at December 17, 2004 10:54 AM

Mary - I'm so terribly sorry for your loss. If my house burned down I wouldn't appreciate someone telling me I should have organized my stuff better and kept the important items offsite. There's always a techie that will immediately take the developer's point of view and that's fine... just don't take that point of view in the blog of the software's victim. NetNewsWire can do a better job... and most likely will as a result of your marketing impact.


Posted by: McD at December 17, 2004 11:35 AM

Brent, I've sent the files. I just want to reassure people here that I'm not writing about this as a victim. I'm writing because I feel that if something isn't working it should be stated as such.

Also, I want notice, not sympathy. I appreciate the couple of people who are supporting me here, but the point is, 1. don't charge for betas, 2. give people appropriate notice at the right moments (because you easily can) of bugs, and 3. help sections are appreciated. Right now NNW is missing a help section, though the beta area of the site asks that users go there for information, and to report or look up bugs. So I can't find out how to fix this myself, which was what I did first, before starting this post.

Also, I know I ultimately lost the data.. but the part that is frustrating is that NNW charged for a beta that shouldn't be charged for... and until I had to pay.. I had redundancy with Shrook, opening it every day to sync it.. but once I paid, I thought I could stop. That is what I mean when I say that NNW give me a cue that the software would be somewhat reliable and at least not lose data. So it is most definitely my own fault. But I don't want others relying on the NNW system as an aggregator of information, with no notice, until this is all fixed.

And the social part? That's about conversing with your customers... notice is what we want. And then we will be happy to put up with bugs and their attendant work arounds.

At this point, though I'm waiting for Brent to respond to the email I sent with files to see what happens.

Posted by: mary hodder at December 17, 2004 11:57 AM

"'s blog was created as a resource to understand the idea of napsterization: the disintermediation by new technologies and digital media of old economy, incumbent institutions and analog frameworks."

Uh, so where's the disintermediation here, and how's it working? Maybe it takes more self-reliance to be disintermediated?

I see that Brent Simmons has made an amazingly calm response, and Mark Fletcher's original offer to recover full RSS feeds for you would (assuming they deliver full content) be as good as it gets and in a preservable, non-proprietary format.  So why not do that rather than all of this speculating that has nothing to do with straight communication between the involved parties?

But I wonder about the wonder of disintermediation. If you happen to use Outlook, Newsgator puts the feeds it aggregates for you into your folders and it doesn't ever delete anything. And it all runs on my machine, with no one else storing my data for me. Even with blogger, which does store all the authoring materials, I have them post via FTP to my site and I back it up back onto my home machine. I would like very much to disintermediate myself from Blogger but I then have to deal with putting software at my hosting service and I have been avoiding that.  But enough about me, how's disintermediation going for you?

PS: OK, I'll speculate too. Did you buy the software or did you subscribe to a service that includes updates to the software for using the service?  It probably doesn't matter with regard to the experience of pain, but it shifts the question about buying betas.  I don't know, I don't use NNW.  Not something I'm proud of, I'm merely calibrating my level of ignorance as to the facts of the matter.

Posted by: orcmid at December 17, 2004 11:59 AM

Christopher wrote:

"The company made the decision to charge for their beta software, not warn users about possible data loss, and now they're reaping the consequences. I'm with Mary on this one: let them suffer."

The company DID warn users about possible data loss. It says right on the beta download page that the software has nasty bugs. The root cause of this problem, IMO, is that someone sent her the software without forwarding those warnings.

I'll agree that it's strange (even wrong?) to sell beta software, but isn't it just as strange to BUY beta software? And then use it for mission critical work?

Christopher wrote:

"As far as backup: She did back it up! She tried to recover the data, and the softtware blew that too!"

First of all, the software can not corrupt her original backup unless she MOVED the backup into the location that NNW reads data from instead of COPYING the data into that location. Secondly, since the beta software had that bug, it was bound to happen again... software is often deterministic like that.

Christopher wrote:

"She can't get it any other way, since it's in a PROPRIETARY format. What, I play tell, should she have done differently?"

It's not in a PROPREITARY format. As far as I know it's in SQLite format, which can be read by a number of tools. SQLite is an open source database library.

It would be nice if everything could be in plain text or XML format, but unfortunately you cannot always get acceptable performance using those formats. I think Ranchero would have an even bigger problem if their software was glacial. It DOES work normally, it does not usually lose data. I've never had problems with NNW losing data, but then I don't think I've used a beta since around 1.0 or 1.0.1.

What should she have done differently? I've already said what, but here we go again:

1) Not used software sent from friends, but instead downloaded the software from the company's web page (where the beta warnings were clear and up front).

2) Not relied on beta software for mission critical data management.

3) Tested her back up before a catastrophe occurred (she was doomed anyway since if the software had a bug in it, it was bound to happen after a restore, hence the reason for not using beta software)

4) Not lost her backup somehow.

I certainly wouldn't have paid for beta software. It's admirable that she did, no question, and expectations were certainly miscommunicated by allowing it to be paid for, but it's still beta.

In all of this, nobody has mentioned that it's possible it wasn't even NNW's fault. A number of things could have occurred that caused the data to become corrupt that have nothing to do with the software or the company or the people behind it.

1) Bad RAM. It's unfortunately quite common, and can cause completely random things to happen on a computer from time to time. Sometimes those random things are harmless, sometimes they aren't.

2) Bad hard drive. If the file was written to a bad area on the hard drive that the operating system hadn't identified as bad yet, data could have been lost.

3) Hardware failures in general. It may sound like a cop-out, but seriously, software runs on hardware and is at its mercy. Hardware does fail! I had an IBM NetVista PC that started rebooting itself half way into starting Windows 2000. I thought it was a virus or a bad drive or bad RAM or something, but it was a freaking capacitor!

An IBM tech came over (since the PC had a 3 year on-site parts and labour warranty, we'll miss you IBM!) and said "see how the top of that capacitor is round instead of flat? It's defective and that's why your computer isn't working properly".

There's NO WAY I would have known this, and for months before I called IBM (I didn't know about the warranty at first) I was blaming Microsoft for this problem (I was sure it was a virus... doh!).

4) System crash. If a system crashes hard, or a user is forced to hard-reset the system, bad things can happen. It's really a crap shoot, if nothing bad happens, you're lucky.

All things considered, I would say her friend shouldn't have sent the software without warning her, the company shouldn't have sold a license to her without warning her, and she shouldn't have used beta software, period, if the data was that valuable.

A bad situation, of course, but all it proves (again) is humans make mistakes, and furthermore, aren't good at accepting that fact.

And to Randy who said:

"Is it acceptable for a Civil Engineer to say that a building will fall down at any given moment, when you least expect it?"

With all due respect, if any Civil Engineer (or Architect) designed a building anywhere NEAR approaching the complexity of a modern PC, operating system with associated worldwide network and software using all of those pieces, I would be shocked if it ever got funding. PCs are obviously amazingly good if people underestimate what a miracle it is they work at all.

I just finished taking a university course on Computer Architecture. Even as a software developer with 10 years of experience I was shocked at just how immensely complex computers are, and I still don't know jack compared to the people (read: huge teams of brilliant people) who design them.

Posted by: Jim Roepcke at December 17, 2004 12:07 PM

Christopher said : "If I was the president of NetNewsWire, here's what I would do: I would send her a check for $5000 for making up for her loss of productivity (she's going to have to redo most of that work). "

I think the point here is that NNW are a tiny company (ISV) and very unlikely to be able to pay a $5000 "fine" for screwing up. There *are* companies that enter into "service level agreements" with customers, whereby they accept financial responsibility if something goes wrong. But these are giants. And, to the best of my knowledge no-one, not even Microsoft, who sells packaged software accepts any financial responsibility over and above the cost of the media (disks) for a failure (go read an MS contract)

As customers, we'd like it to be the case that vendors accepted the risk of bugs losing our data, but it's not the way the industry works (and probably it couldn't work).

This is why Mary seems to be so unfair.

When things went wrong, she was offered what's the standard "appology" and recognition that NNW had fucked-up : a refund. And because she was pissed-off she called the people who offered it "morons". In public.

But they weren't morons. They were just ordinary developers working pretty much the way all ISVs do. They have bugs. They make mistakes. They have break-downs in communication. (Maybe some people never sell "beta", although I've never heard the "rule" that "if you charge it isn't beta".)

They did as much as you can expect any ISV to do. Refund + try to help. That's not moronic. That's what you do have a right to expect. But it's not clear you have rights over and above this.

If you expect SLAs, go buy a million dollar contract to manage your data from IBM.

Posted by: phil jones at December 17, 2004 12:09 PM

FYI, I was just doing some reading about NNW 2.0 beta (I'm on the beta list, despite not being an active beta tester :-(), and it seems it isn't use sqlite anymore. I read that the history is split up into several XML files. (It sounds like that's what the FeedData folder contains)

Huh. So, it's even less propreitary than I thought!

(Doesn't change anything, but I didn't want to misinform anyone)

Posted by: Jim Roepcke at December 17, 2004 12:20 PM

Mary writes: "Also, I want notice, not sympathy. I appreciate the couple of people who are supporting me here, but the point is, 1. don't charge for betas"

Why did you buy a beta?

"That is what I mean when I say that NNW give me a cue that the software would be somewhat reliable and at least not lose data."

They could do that with known, fixed bugs, or known unfixed bugs, but that's not possible with unknown bugs.

I also frankly suspect you are abusing the notion of aggregation a bit, and defining it in a way that is not the definition which aggregator developers use.

It's an aggregator of ephemeral feeds, not an aggregator of persistent data. It's like the difference between a TiVO and the historical tape archive at CNN. You're trying to use a TiVO as an archival database.

Posted by: Jon H at December 17, 2004 12:27 PM

Gee, why am I not surprised that Winer's sycophants like Rogers Cadenhead are rushing to the defense of Brent, one of Dave's former associates? I guess that's why Dave blogged this, so his buddy-buddies could pile on.

Posted by: Anon at December 17, 2004 12:31 PM

The problem being anon, that DW will *always* take the user's side in these types of discussions (rightly or wrongly), and given Dave's latest post on the matter makes your point a little, erm, blunt.

Posted by: Ross at December 17, 2004 01:00 PM

I thought Rogers' comment was off the wall. I cringed as I read it. I don't think he would be so understanding if he were the cusomter and the vendor was his utility or phone company. Why should software companies, no matter how small, want to provide *less* service to their customers?

I thought Brent's response was great, even if it was a little late in coming (maybe he was working on something else). Validate that there's a problem, and get to work solving it, the best you can. If you can't you can't, but you don't know that until you try.

And I think the hero of this is Mary. It takes guts to stand up to all of you. But I know Mary to be highly principled and courageous, so I decided to point here, to see what would happen.

To those who think she's wrong, I think you need to go to a user's conference and listen to how people use this stuff. It'll change your perspective.

Posted by: Dave Winer at December 17, 2004 01:04 PM

"Gee, why am I not surprised that Winer's sycophants like Rogers Cadenhead are rushing to the defense of Brent, one of Dave's former associates? I guess that's why Dave blogged this, so his buddy-buddies could pile on."

Obviously, Dave says that Mary seems to be right, and so Rogers sycophantically chimes in that Mary caused the situation by using software of unknown provenance. Yeah, that's what I call rushing in as a buddy-buddy.

Jim's absolutely right -- why in the world would anyone install beta software they got third hand, apparently not follow along with the beta forum, and then complain that it's all the horrible company's fault.

Posted by: Brian Carnell at December 17, 2004 01:05 PM

Ross you don't quite get me. I will tend to take the user's side, because if the customer is wrong, what kind of business do you have?

I first learned this when CEO of Living Videotext, coming across a group of engineers in the company kitchen telling stories of how dumb some users are.

I stopped them, and said they're wrong, our customers are the smartest people on the planet. They sorted through all the options, all the compeititon, and chose the very best product in the world.

In other words, if our customers are stupid, then they probably chose the wrong product. No company gets very far thinking ill of their customers because it means they don't think much of themselves.

Note that I said Brent did a good job. But I think the others here, starting with the first comment, need to look at whether they made the customer feel stupid. I thought the first comment, saying he'd never piss off Mary, was a bit on the arrogant side. He's always telling people to use his product as the solution to every problem known to man. It's like chicken soup, cures all ills. I cheered when Mary told him she was familiar with is product, and didn't think it would cure her problem.

You guys finally met a smart and feisty user. I hope you meet many more. When you can make them happy, you will truly have a product worthy of users and awards.

Posted by: Dave Winer at December 17, 2004 01:10 PM

Oh I think I understood Dave, I agree totally that the customer is *very nearly* always right. I go to excruciating lengths to make sure my users are happy (which given that it is used in a call centre is not as easy as it might sound). I've also had my fair share of time working with team members who have nothing but disdain for users.

Unfortunately sometimes users *are* wrong. If you were an architect and the user demanded that you build a house with no foundations, I'd hope you would refuse to do it. I know I would and have, well you know, not refuse to build a house, but .. whatever, you get my point.

And if there were a developers contract on how they would treat and interact with users, there should most definitely be a contract on how users will interact with developers. As I mentioned earlier, if the developer is sensible enough to make my data open, then the user should take some effort in ensuring that they don't recklessly lose that data if it is that important to them.

This whole thing could have been done without resorting to calling developers morons, in fact it could probably have been done without naming names at all and *still* got Brent to help solve the problem. My only real problem with this post is that it lays the blame for the situation squarely at one party's door when there are obviously mistakes on all sides, and it seems Mary wants to use the brute force approach of bringing it all out into public in an apparent attempt to shame the developer into fixing something (and I apologise in advance if this is not the case - just my perception).

Posted by: Ross at December 17, 2004 01:21 PM

Mary -

Loosing data is always tramatic... It has the same emotional vibe to getting robbed. A better analogy might be a housefire where you loose personal items that can never be replaced. It makes no sense to come into the shell of a burned out house and suggest that personal items should have been stored off-site.

I'm sorry for your loss and pain.

All the lessons from here are object lessons and
you should grieve or rant as you see fit. Especially, here in your personal space. Most of the A-list bloggers dont' expose themselves to
comments amymore because everyone has an opinion and few care about you're feelings. They also don't want to give free rides to someone with an agenda... "Let them use their own blog for pissing me off".

So, tomorrow is another day... yada, yada.

(Having said that I'm eager to jump into the Users
vs small developers issues... but... another time and place. Name calling, hot passions, and agendas
have no place at the burned home.)

PS - I love your writing... I'm Bookmarking you.
Enjoy the expanded audience that controversy and Dave Winer can bring.


Posted by: McD at December 17, 2004 01:57 PM

Hmm, apparently my very first comment is being misinterpreted. I never wanted it to be an advertisement for anything; I only wanted to offer whatever help we could. That, and a lame attempt at humor.

Posted by: Mark Fletcher at December 17, 2004 03:23 PM

I think Mary would have received a lot more sympathy had she not been so incendiary.

Did I see someone comparing the loss of some RSS feeds to a house burning down and a family dying?

I think it's OK to charge for beta software. If you don't want to pay then don't use it. And I don't think payment guarantees a release-quality product.

Even though you did not obtain the software from the developer's web site, it might have been a good idea to visit the site to get some information about the product, especially if you were expecting use the software rather heavily.

It makes sense for different pieces of software to have different degrees of quality/reliability/etc. The development of an RSS reader should not be as rigorous as the development of NASA software (which even has critical bugs on occasion).

I'm surprised Mary complains about the "social media" aspects of this situation given unprofessional demeanor expressed in the original post.

Posted by: pb at December 17, 2004 04:14 PM

A) Its beta.
B) All software can crash and lose stuff.
C) Shrook is so freakin buggy i just gave up on it. When 2.0 came out i hated the interface so did not even try to learn it.
D) Netnewswire rules
E) sorry to your friend but stuff happens.

Posted by: me at December 17, 2004 04:14 PM

Wow, that sucks.

Maybe I'm too huge of a nerd, but my suggestion for storing backups of newsfeeds would be to use something like MagpieRSS (a PHP RSS aggregator) and tie it into a MySQL database. That way not only do you have everything stored in a database, but you can leverage the querying power to search / analyze / compare / manipulate the data later.

I'd totally be interested in collaborating with some folks on this is there is a demand. I'm a comp sci student / php stud-muffin and could really use some more crap in my code portfolio. Let me know if you're interested: nick {at} deceptykhan{.}com! Cheers!

Posted by: nick b. at December 17, 2004 04:26 PM

Sung to the tune of the original Batman song from the campy TV show:

Beta beta beta beta BETA! BETA! BETA!
repeat until exhausted.

I understand the frustration. But the beta warnings were really clear when you downloaded it from Ranchero. This software could explode. Please keep innocent bystanders far from the blast site.

Posted by: Glenn Fleishman at December 17, 2004 04:26 PM

One more thing: people rarely back up even critical things much less their RSS feeds.

Posted by: pb at December 17, 2004 04:33 PM

Mary, you said "My point is not that the software is beta or has bugs... my point is that bugs they know about before they make the sale should be told to the buyer."

Jesus Christ. His whole fucking bug database is online. For every application Ranchero makes.

It's not like he was hiding the bugs from you. He's more open about bugs than practically ANY other software developer out there. You're insane to be ranting like this. Completely nuts.

Posted by: Ben Stanfield at December 17, 2004 04:34 PM

Hi Mary, it's my personal mission to convince people to STOP TRUSTING OTHERS with the integrity of their data - when I think I might be listened to : )

I saw Dave Winer's comment from Bloglines and I followed the link from there.

first off let me say it takes a LOT of guts to have this conversation out in public.

Plenty of "A-list" bloggers don't and in my opinion lose out on an easy opportunity to show their character. Others DO allow comments and show their bad character by deleting relevent posts, making poor arguments and changing the subject / not standing by their words.

u rock, they suck.

Speaking as someone who has lost, perhaps, 3 full systems (docs, installs, prefs) in my first 15 years of computing, I finally learned that you can not trust ANY VENDOR with your data.

Not IBM, Western Digital, Seagate, Apple, Microsoft, Dantz, Adobe, Symantec. None.

Hardware, software, operating system, utility, application. None.

It doesn't really matter who is at fault. One day somebody is going to let you down. Some might blow you off, some might try to help you. Doen't matter. Because it's not their data, it's yours.

Complain about these guys if you must, but sooner or later this will happen to you again.

On a computer with important work on it, anything less than daily methodical backups is just asking for trouble and the only person you will have to blame the next time something happens like this will be yourself.

Please take my advice and set up some form of daily, reliable and redundant backup.

Since I did the research and implemented my system, I have not lost more than 1 day of work in 5 years.

I'm a visual artist so I generate gigs of data weekly. Not everyone will need a system like mine (Mac, DAT tapes, incremental nightly backups) but everyone NEEDS some system.

Unlike what Dave Winer says about the commenters here, I don't think you are in the wrong. I do think that this will happen to you again if you don't take steps to prevent it.

my 2 cents


Posted by: willc2 at December 17, 2004 04:38 PM


Again, I sympathize with your frustration at having lost data and work. If I understand the technology involved you probably can't "backup" all the files in a distributed architecture but rather reply on NetNewsWire to some degree and it failed to behave reliably... I've tried to glue back together an application and it's metatdata on occasional and had to give up. Just getting back some snapshots of the files might not do it if "timestamps" are involved and the application says "timeout".

Anyway, more object lessons. and more traffic for
your site. and more advice. when advice is probably NOT what's needed. The "professional"
communication style some are requesting makes a lot of sense in the business world but single users should be allowed the freedom to use their
natural voice which may not contain language suitable for children or software developers
(who know how incredibly fucking hard it is to write perfect code... and satisfy the needs of all users... oops, that's not helping is it?)

Try to enjoy the stimulating comments and keep
pushing for better tools.


Posted by: McD at December 17, 2004 05:55 PM

"I don't think he would be so understanding if he were the cusomter and the vendor was his utility or phone company."

If we're having a narrow discussion about Mary's predicament and Ranchero's response, of course it's unacceptable that the software ate her data and she should expect better performance. The customer is always right. Her work is important. Commercial software should reliably do the job you bought it to do.

But if we're expanding the discussion to look at the situation as if we are a bunch of software developers talking shop, I don't see much more Ranchero could have done to provide forewarning that data loss was possible.

She suggested that a pop-up dialog during the install should have warned her about this bug. I think any developer who believes a bug is serious enough to merit such a dialog should be working on a fix, not a warning, because many if not all users zoom past warnings.

All I'm trying to say here is that everyone makes shitty software, and users are better off counting on that and acting accordingly.

I can't count the number of times I've lost important work in Microsoft Word over the years. More than a decade of development by the richest and largest software company on the planet, and I'm still hitting Alt-F-S every 30 seconds to mitigate disaster.

Posted by: Rogers Cadenhead at December 17, 2004 07:09 PM

Gee, why am I not surprised that Winer edited the remarks on his blog to say something entirely different than when I originally commented? Now his pointer to this discussion comes off comes off as sympathetic to Mary's plight, which it definitely did NOT before he edited it. And of course, he'll deny he made any such edit.
Come on people, you know what a deceitful asshole Winer is.

Posted by: Anon at December 17, 2004 07:26 PM

PB asks "Did I see someone comparing the loss of some RSS feeds to a house burning down and a family dying?"

Yes, (housefire)... calling upon the power of analogy to try to
see someone else's point of view (not sure about the
"family dying".

That's my point... have some sympathy and move the
"user vs software developer" discussion to another
forum. Mary doesn't care if we think her feelings and rage are "out of line" and she shouldn't.

Like many have said... object lessons are the result of loss. She's a smart person and will
learn but the pain is no less. Many have experienced the same and can relate.


Posted by: at December 17, 2004 07:49 PM

But don't you think that if she approached Brent the way she did here that he wouldn't be too excited to help her out? It's not real satisfying serving customers throwing around the f word and calling people morons.

The burning house and family dying analogies are a stretch in this situation. Better is that the paper boy missed your house this morning.

Posted by: pb at December 17, 2004 08:20 PM

Mary sent me her data. Here's what I found:

Mary has NetNewsWire set to keep news items for 999 days. In about 10% of her feeds, old items -- items that have disappeared from the actual feeds -- were deleted from NetNewsWire instead of kept. All but one of these feeds were from the same service.

Unfortunately, I can't recover those old items without a backup from before the items were deleted.

Here's a copy of the email I sent to Mary:


Hi Mary,

Thanks for sending me your NetNewsWire prefs and data. Even without also seeing your subscriptions list backups, we were able to identify a bug which could account for the loss of those news items.

When NetNewsWire deletes items, it checks the date the item was published, as specified in the feed.

My theory is that sometime recently there was a glitch, probably temporary, with Technorati watchlists, and news items in those feeds got some incorrect old date. NetNewsWire then deleted those items because it thought they were old.

This bug is easy to fix: NetNewsWire also stores the date it first saw an item, and will use that date if it's more recent than the feed-provided date.

We will fix this bug and release a new beta. We will also continue to work on the safety of NetNewsWire users' data. It's an ongoing process.

About recovering your data...

I'm very sorry that I can't recover the old items from the data you sent me. If you have backups of your FeedData folder from before this happened, then I could do it. If you do, then please let me know, and I'll be happy to recover the old items from the backup.


Posted by: Brent Simmons at December 17, 2004 08:37 PM

"Mary doesn't care if we think her feelings and rage are "out of line" and she shouldn't."

Why shouldn't she? If the rule for companies is "the customer is always right" isn't the rule for blogs "the reader comment is always worth attending to" ;-)

Posted by: phil jones at December 18, 2004 05:07 AM

Great job Brent. Sounds to me like we should all be talking about Technorati's bugs. I mean in actuality the data provided by the feed should be authoritative. Keep making the great software!

Posted by: Ian Landsman at December 18, 2004 07:43 AM

Nick B: the project you describe (a Magpie-based aggregator that stores the data in MySQL) exists already: Feed on Feeds, written by Steve Minutillo. It's just awesome -- I've been using it as my only RSS reader for the past year, and I have a MySQL database that's a testament to that. I'm sure Steve wouldn't mind having you hop aboard the Sourceforge project, though!

(And sorry, but I'm nearly speechless at the notion that Dave Winer came into a thread and said that companies should realize that customers are always right. The experience that the Fortune 50 company I worked for had with Userland -- when Dave was at its helm -- is well-documented and horrid; despite being one of the only customers that explicitly paid for their highest-level support option, we were publicly ripped, then ignored, and then ultimately told to find another product because they wouldn't support us anymore. Sorta makes Dave's arguments in this thread mean a lot less to me, and I'd assume to anyone else who watched that episode unfold back in 2002. I only bring this up as a reminder that people's comments, especially in pile-on threads like this, should be viewed through the lens of other things they've said, done, advocated, and whatnot. The truth is sometimes a bit different than you'd think...)

Posted by: Jason at December 18, 2004 08:32 AM

So Dave, how right were your customers when you closed up shop on with no notice, no backups, and thousands of pissed-off users??,1284,63856,00.html

Just another example of your hypocritical ways... "The user is always right, except when they are MY users!" Muhahaahahahaaaaa!!

BTW, WTF pays for beta software?!?! That's just whack!

Posted by: WeinerBoyz at December 18, 2004 10:46 AM

I wouldn't attach my name to a comment that uninformed either, WeinerBoyz.

Two days after Weblogs.Com went offline, Dave was working with me to bring them back up on Buzzword.Com. I haven't heard from a single customer of his free hosting service who is unhappy with how all of that went down, and at this point, they've all had their data for months and can be completely candid with how they feel about it.

Regardless of how it was reported in the press, it was a simple server outage, and Dave worked harder than anyone to restore service.

Posted by: Rogers Cadenhead at December 18, 2004 07:16 PM

Just a note about the use case.

Mary's power use may be leading edge, but it is well within scope. If you spend hours of each workday online or at your laptop, then any RSS client becomes integral. It replaces bookmarks. It replaces addressbook entries. It replaces listerv archives. It becomes the place you look for your own blog posts, a backup brain. To pick up threads of conversation. To detect trends and be a little smarter. The place to find contact info, places to go, things to do.

So it's indispensable.

But it must scale too.

About the assumptions: disk is cheap (comparatively) so there's no cost to adding feeds and saving posts. Six months after I started adding feeds, I had 500. Six months later, more than a thousand. Ask Scoble how many he has these days. With time, and exploding feed availability, it will be nothing to pick up 20-100 a day. That's thousands a year, hundreds of thousands of posts before you know it. All reflecting your social and reading and writing behavior, your interests and values.

Quite an asset, unique in all the world. And personal. (Something you might be upset about losing.)

So don't scold Mary for pushing the envelope and holding the lovely developers accountable. It's in all of our interests. Because someday sooner than later, we'll all be in her shoes.

- phil

Posted by: Phil Wolff at December 18, 2004 07:25 PM

Guys, Can you please take your fights to someone else's blog. I'm shutting off comments on this post because I'm not interested in hosting some sort of off topic flame war. I just don't care and that shouldn't be happening here. Thanks.

In fact, upon thinking about it, I'm deleting the last two comments because they have nothing to do with anything here. They belong somewhere else.

Posted by: mary hodder at December 19, 2004 06:54 AM