Reindl Harald wrote: > schrieb Bob Proulx: > > Being able to undeliver spam after it has been detected later and if > > it is as yet unread is none of those bad things. This is a positive > > anti-spam feature in the core feature set of an email provider. > > honestly i would not want to get a message removed which was already > in my inbox because someone *later* decides it is spam and hence > the sender got no NDR in case that action was a false postivie
You haven't seen the message yet. You never knew it was ever in your inbox. It doesn't move messages you have already seen and read. It only does this if you haven't seen it. Since you haven't read it yet you never knew that it wasn't delivered to the Junk folder on the first pass. This is identical to the behavior you will get if SpamAssassin were to have identified it as spam on the first pass. It is impossible for you to tell the difference. A difference which is impossible to tell is no difference. There is never an NDR (non-delivery receipt) sent for email classified as spam. Or at least shouldn't ever be one sent. That would create backscatter spam itself. So again that is no difference. The only time an NDR should be sent is if the email message is identified as spam and rejected at SMTP time. Many sites have no ability to do this and instead accept the email and filter it later. If it is identified as spam later it cannot/should-not generate a bounce later or it will be backscatter. > in case of a false positive the sender can pretend with > his logs that my server accepted the message and he is > right in pretend that - so any message which was accepted > with "250 OK" has to made it in my inbox and there is no > but and if in that context > > that it what makes mail *relieable* Sorry but that did not parse. I think you are saying that once accepted at SMTP time email must be delivered. Which is obviously not true once anti-spam is applied. Spammers would love it though. > * no sending attempt not confirmed by "250 OK" is counted as > successful and retried up to 5 days > * any message which is rejected produces a NDR from the sending > server to his user (or ignored by a spammer) > * any message which can not be successful delivered > produces a bounce from the sending server to his > user (or ignored by a spammer) > * any message not producing a NDR within 5 days can > be counted as delivered None of the above is related to Googles ability to reclassify messages later. > anything which leads in > > * accept and drop silent > * accept and reject internally leading in a backscatter > > makes email at a whole unrelieable and so does the same harm as spammers > you can't justify with the fight against spam any bad design This conversation is taking a turn off into the weeds. I can see that this is some crusade topic of yours. Sorry but I am not interested in investing the time to delve into it. > if you do you end where the USA ended with justify anything with the > fight against terrorism - that's really the same: the other side won > because one did enough damage to itself to make them win I see that we have left the road far behind, driven over the weeds along the edge, and have gone seriously off-roading and off topic. This has nothing to do with the email topic we started discussing. > > Therefore the simple argument of "more code bad" does not apply. > > Otherwise everyone who starts a program by copying "Hello world." and > > expanding it would be stopped immediately by the inability to add code > > in order to have it provide more functionality > > you need always to draw a line - but with care > feature creep in many cases proved later "OK, now we can throw > away all that code and start from scratch because it became > unmiantainable over the time" and i am somehow tired of all > that rewrites starting each time with the early bugs again I haven't looked at Google's code base for undelivering mail later classified as spam. If you haven't either then the assumption that it *must* be bad from the start is unfounded. And even if their version 1.0 were a mess then surely all would agree that Google easily has the resources to rewrite it several times. Bob
