Move SpamSieve to another computer

Hi, I’m new with SpamSieve and my experience with it has only been a couple months. But I found its very effective in filtering spam mails.

I need help in how to move SpamSieve from one computer to another computer as I’ve replaced my old desktop with a new one. How do I transfer all what SpamSieve learned in terms of Spam to the new computer? What files do I move and/or export and copy? Thanks in advance.

Please see this page.

Thanks for your quick reply.

This should really be an explicit function.
I just downloaded SpamSieve, and this was the first question I had. Spam filters are nearly useless if you can’t copy them between computers, so I wasn’t about to bother using the product without this ability.

While copying the files you mentioned isn’t hard for computer-literate users, it’s not really what you’d call a release-quality UI either. I hope you guys plan to add explicit import/export menu options for filters. Not only is this more user-friendly, but it’s more robust. Someday you may change the storage format, filename, or location of the filters.

Could you explain a bit about why you think this and what you’re trying to accomplish? I don’t expect many SpamSieve users to need to copy the training data between computers. If you get a new Mac, everything will automatically be copied over by Apple’s Migration Assistant. If the Macs are used by different people, it’s not advisable to copy the training data because each user should be training it with his own messages. And under normal use, there won’t be any manually created rules that need to be transferred.

Sure. It’s because people today frequently have two or more computers. Perhaps an iMac and a laptop. Your main computer will accumulate filtering data far more quickly and comprehensively than a secondary computer, especially if you use it all day long. Users shouldn’t have to train SpamSieve all over again on another machine. For example, if I take my laptop on a trip, I should be able to update its spam filters easily before leaving.

So it’s a problem with just two computers. I have three: a Mac Pro at work that handles the bulk of my mail, an iMac at home, and a laptop. Am I supposed to train all three of these separately?

Also, when I asked this question, I was under the impression that you could construct specific searches for patterns and keywords in mail content. I’m disappointed to find that this isn’t true (because I’m willing to do that work to avoid false positives), but I’m giving SpamSieve a shot anyway.

Well, I’m not still not convinced that some sort of import/export feature is the right solution here. I think training is really easy, takes just couple minutes, and it’s not critical that all the Macs be perfectly in sync. You’ll get very good accuracy, anyway. For the average user, it’ll be easier to just train SpamSieve separately than to figure out how to use a network or USB stick to transfer the files.

Secondly, if you have two Macs that take turns being your primary Mac, I would imagine that you already have some sort of synchronization software to handle your documents. If so, you could simply include SpamSieve’s data folder in that sync. An import/export feature, on the other hand, would require that you manually go into SpamSieve to do this.

Are they checking the same mail accounts?

You can do this using SpamSieve’s whitelist. What I meant was that most users won’t ever need to manually create any whitelist or blocklist rules because SpamSieve automatically learns from the messages that you receive.

And yet how to do so is listed as a frequently asked question in the manual.

I don’t know where I said that two computers take turns being the primary one. In fact, I emphasized that one computer is the primary computer and others were secondary, leading to much better filtering data on one computer.

I wrote a shell script to back up and restore preference files to a thumbdrive. This is not something most users are going to tackle.

Yes.

And thanks for the info about the whitelist, but as far as I can tell, it’s a… whitelist, stating criteria to always allow messages through. Probably 90 percent of my spam comes from about three sources, all of which have easily identifiable patterns in the raw source of the message (which Mail’s built-in filters foolishly don’t allow you to search). If I could, for example, search for URLs ending in “ru” or “cn”, that’d eliminate tons of spam with very little chance of false positives.

It’s in the FAQ because some more technical users want to do this and can easily do so based on the written instructions. The average user doesn’t read the FAQ and doesn’t know how to transfer files between Macs. For such a user, it is both intuitive and easy to setup SpamSieve separately on the different Macs. Perhaps in the future something like MobileMe syncing would be even better, i.e. just click a preference and it works automatically. A manual import/export feature is neither here nor there in my opinion: not much easier for the technical folk, and still out of reach of my mom.

On your trip, you don’t have access to your Mac Pro, so you are using the laptop as your primary Mac. People in this situation often sync their files between the two Macs. In contrast, some people use different Macs for different kinds of tasks, e.g. personal and work. In these cases, the Macs may contain distinct files that are not synchronized.

By the way, if one of your desktop Macs is usually running, you could set it up as a drone, and then you wouldn’t need to install SpamSieve on the other Macs at all.

Yes. You said that you wanted to avoid false positives. That’s what the whitelist does. There is also a blocklist, but creating blocklist rules can only increase the number of false positives.

I don’t recommend creating rules like this. Mostly, you’ll just catch messages that SpamSieve would have caught anyway. And your rule may inadvertently match a good message that SpamSieve never would have thought was spam. I realize that you probably have in mind rules that you don’t think would ever lead to false positives. But my experience is that it’s surprisingly hard to anticipate which messages a rule might match.

So my general advice would be not to try to micro-manage SpamSieve, at least not at the beginning. Just train it according to the guidelines and let it do its thing.

I’m sure some people would appreciate this, but this assumes that everyone who would use the feature is paying $100 a year for a MobileMe account. That’s more convenient for some, but a stand-alone solution is universal. Now if the learned filters could be kept on MobileMe OR your own FTP/WebDAV server (I use the Xmarks bookmark-syncing plug-in like this), that would expand the options for knowledgeable users quite a bit.

I keep everything on a couple of external drives, so it doesn’t matter which Mac I’m using for which purpose. No syncing needed; I just run regular backups.

Good suggestion. I considered doing this with the boatload of filters I set up in Mail, but I’m just not willing to run a Mac Pro or even an iMac the whole time to filter spam.

I understand. I’m worried about false positives because I don’t know the criteria SpamSieve is using and how they’re weighted. If I set up every rule by hand, I know exactly what it’s doing. I have a massive collection of rules in Mail, and the false-positive rate appears to be nearly nil. But admittedly there is some risk, so maybe I won that airplane but never found out about it…

I’m going to do that. I’m not assuming the product won’t work; don’t get me wrong. And its self-management does reduce the importance of the import/export feature. I’ll just add the relevant files to my backup/restore script.

Thanks for the info!

Well, storage is the easy part. The hard part is writing a record-level sync engine. Apple has already done that and built it into the OS, but currently it only works with MobileMe.

I suggest disabling those when using SpamSieve. This will make it easier for you to see which messages SpamSieve has let through, so that you can train it properly.

Yep. I did follow the directions, so I turned off all my other filters and Mail’s junk detection.

While SpamSieve did start filtering some things effectively, this morning I came in to find numerous messages with obviously spammy “from” strings. Some examples:

Try Viagra for Free
Ab Circle As Seen on TV
VIAGRA© X Official Site
Rated #1 Superfood on Oprah

This was an uncaught subject:
What if you could save up 92% OFF your printer ink?

And four false positives.

Anyway, I’ll keep training it…

Please send in your log file.

Log file & false negatives sent.

I have not received them. Please re-send.

OK, sent them again. With some more false negatives from today.

I’m still not seeing it. Please send just the log, and compress it to avoid tripping a server-side spam filter, to spamsieve-fn@c-command.com.

I did compress the false-negatives collection, and sent the log file plain. If your spam filters are rejecting that, you need to fix them pronto. I followed the instructions on your Web site for submitting this information, and if you didn’t get it, it means other users aren’t able to help you improve your product either.

I have now re-sent this information three more times, with different strategies:

  1. Log file as a zip file, false negatives as a zip file.
  2. Log file as a zip file, with a link to download the false-negatives file.
  3. Links to download both the log file and the false-negatives file.

All were sent to the address you provided, from the same system with the same config used to send the others. Please fix your spam filters accordingly.

There are no spam filters on my end. However, there may be one between your mail server and mine that out of both of our controls.

Well, in seven years I think you’re only the third person who reported sending a log file that didn’t get to me.

I received (2) and (3). Thanks!

It looks like about half of the false negatives were because the sender was in your address book (and was your own address). There are some suggestions for dealing with that situation here. I see only three false positives in the log, none in the last day, so I don’t expect those to be a problem going forward.

Is SpamSieve not considering the textual “name” that’s included for the originator? Because although the fake originating E-mail address is my own, the “from” name is “Viagra X Official Site”.

If it’s ignoring that text, it’s never going to be as effective as it should be. About half the spam I get has highly descriptive “from” names that include clear indicators of spam.