RE: Relation email - person (re: Mail sent to midcom)
|-Original Message- |From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] |Sent: Thursday, February 15, 2001 3:49 AM |Subject: Re: Relation email - person (re: Mail sent to midcom) | |25.00% defunct | 0.1% duplicates (same person, different addresses) | 0.01% wrong person | |which is a pretty strong evidence of Harald's assertion: | ||The mapping address - person is pretty strong, and mostly single-valued. ||The mapping person - address is multivalued, and getting more so. | |One would expect that in "clean" data, these mappings would |be even stronger. The first and second statistics can be taken care of with management. The last one is of concern but could also be taken care of with management. Not sure that it is strong evidence. I have multiple e-mail addresses, some of them redirections to other addresses and others that map finally through redirections to multiple addresses and individuals. Take mailing list addresses for instance where a single address resolves out to multiple individuals, some in fact may not be to individuals but expanded out in other directions, add in wap and it starts getting complicated. It may be desirable to have an authoritive address for each individual and I assume this is where this thread is heading. I'm interested in the subject of e-mail which is why I broke my lurking :). Darryl (Dassa) Lynch.
Re: Relation email - person (re: Mail sent to midcom)
Vernon Schryver wrote: It's hard to know when a username is truely defunct. Depends on the corporation. At Netscape, we had an LDAP server that ruled everything: email, NT and NFS fileservers, phones, and key cards. When someone left the company, HR updated the LDAP server, and that username was gone *everywhere*. -- /==\ |John Stracke| http://www.ecal.com |My opinions are my own.| |Chief Scientist |=| |eCal Corp. |The plural of mongoose is polygoose. | |[EMAIL PROTECTED]| | \==/
Re: Relation email - person (re: Mail sent to midcom)
From: John Stracke [EMAIL PROTECTED] It's hard to know when a username is truely defunct. Depends on the corporation. At Netscape, we had an LDAP server that ruled everything: email, NT and NFS fileservers, phones, and key cards. When someone left the company, HR updated the LDAP server, and that username was gone *everywhere*. The use of LDAP or any other technical mechanism is an indicator and not a determinator of when a username is truly defunct, because the death of a username is the result of a non-technical decision. Recall the point concerns whether the the mapping of username-person is close well defined in the mathematical sense. If Netscape was as reluctant to re-issue usernames as most outfits, then it counts as one that had trouble knowing when a username was truly defunct, and so helped keep the (username,person) mapping well defined. I somehow doubt that Netscape's RCS or other source control archives were rewritten to remove the references to old usernames. I bet that I could list a dozen usernames that could never have been re-issued to engineers at Netscape. While those usernames might be turned off via LDAP, they probably could never be made truly defunct. (Never mind that I suspect Netscape had plenty of people who ran their own /etc/aliases and /etc/passwd files that were not disabled by any central LDAP servers. Judging from their private words to outsiders, some of those people were not exactly impressed or thrilled by the activities of the network administrators at Netscape.) Vernon Schryver[EMAIL PROTECTED]
Re: Relation email - person (re: Mail sent to midcom)
At 17:18 13/02/2001 -0500, Keith Moore wrote: I also wonder about Harald's sample - might this particular group of people be more likely to - understand the value of a stable email address - pick a ISP that provides good service and has good potential for longevity - have his/her own personal domain name - forward his/her mail from older addresses to newer ones. than the average email user? with 4.400 Yahoo accounts and 7.400 Hotmail accounts in the mix, I doubt that it is completey untypical .-) -- Harald Tveit Alvestrand, [EMAIL PROTECTED] +47 41 44 29 94 Personal email: [EMAIL PROTECTED]
RE: Relation email - person (re: Mail sent to midcom)
|-Original Message- |From: Harald Alvestrand [mailto:[EMAIL PROTECTED]] |Sent: Wednesday, February 14, 2001 5:41 AM |To: Mike O'Dell; [EMAIL PROTECTED] |Subject: Relation email - person (re: Mail sent to midcom) | |I recently had the dubious pleasure of sending out 40.000 |emails to a set of email addresses gathered (with the owners' approval!) over |a period of seven years. | |The result was roughly 10.000 bounces (naturally), dozens of |requests to merge multiple registrations for the same person, and on the |order of FIVE occurences of an email address previously used by one person |now being used by another. | |The mapping address - person is pretty strong, and mostly single-valued. |The mapping person - address is multivalued, and getting more so. | |Not quite "not working", if we take it for what it is. I would consider such results the fault of the list maintainer and not a fault in the email system. Much like physical addresses used within the postal system, anyone maintaining a list needs to provide a means to maintain the validity of the data. If the data is invalid it is a cost the person using the data has to carry. It doesn't mean that all the data is invalid, just the means to keep it current was inadequate. Most mailing lists for instance employ means to maintain the integrity of the subscribtions, including regular probes. There are means available for other types of lists, a lot depending on the usage and value. Darryl (Dassa) Lynch.
Re: Relation email - person (re: Mail sent to midcom)
Harald Alvestrand wrote: The mapping address - person is pretty strong, and mostly single-valued. Interesting. Hypothesis: this might happen because (a) ISPs (in the US) try to avoid reusing addresses in order to avoid ECPA problems; and (b) corporations try to avoid reusing addresses because they'd rather have email bounce than have confidential information go to the wrong person. -- /==\ |John Stracke| http://www.ecal.com |My opinions are my own.| |Chief Scientist |=| |eCal Corp. |Campbell's has it wrong--it's "Never | |[EMAIL PROTECTED]|underestimate the power of *chocolate*". | \==/
Re: Relation email - person (re: Mail sent to midcom)
Hypothesis: this might happen because (a) ISPs (in the US) try to avoid reusing addresses in order to avoid ECPA problems; and (b) corporations try to avoid reusing addresses because they'd rather have email bounce than have confidential information go to the wrong person. I also wonder about Harald's sample - might this particular group of people be more likely to - understand the value of a stable email address - pick a ISP that provides good service and has good potential for longevity - have his/her own personal domain name - forward his/her mail from older addresses to newer ones. than the average email user? (though presumably IETF folks are also more likely to fit the above criteria than the average email user) Keith
Re: Relation email - person (re: Mail sent to midcom)
From: John Stracke [EMAIL PROTECTED] The mapping address - person is pretty strong, and mostly single-valued. Interesting. Hypothesis: this might happen because (a) ISPs (in the US) try to avoid reusing addresses in order to avoid ECPA problems; and (b) corporations try to avoid reusing addresses because they'd rather have email bounce than have confidential information go to the wrong person. To the contrary is the corollary to "never attribute to malice that which can be explained by stupidity." The corollary goes something like "never attribute to careful planning or reviews by lawyers (e.g. privacy issues) that which can be explained by laziness." It's hard to know when a username is truely defunct. If you recycle a username too soon you cause grief, from mail going to the wrong place to not forwarding ex-employee mail (many outfits still do) to not being able to accommodate a customer that left inadvertently (e.g. failed to pay a bill) to minimizing hassles with false spam complaints. You must wait at least a few weeks and probably at least 3 months before recycling a username. That implies that the lazy tactic of rarely or never recycling usernames is best. And that has implications for the (address-person) relation. Note also that the "dictionary attack" spammers are trying hard to teach people to pick usernames that are globally unique. (Those are the spammers with lists of several 100 and perhaps 1000's of common usernames that they mix with their lists of domains. Never mind that their lists make dandy spam traps, addresses that trap SMTP bodies that are rejected when later sent to real addreses.) Vernon Schryver[EMAIL PROTECTED]