dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
9
share rss forum feed

InvalidError

join:2008-02-03
kudos:5
reply to JMJimmy

Re: Proud of TSI for it's actions in Jan 14th hearing

said by JMJimmy:

they gave the court a major point to consider: Only ~50% of the IPs could be associated. That has huge legal significance because this is a civil case and the odds are important.

The reason(s) why 50% of IPs could not be resolved to an account are important too and may not necessarily have any negative impact on the accuracy of the other 50% that were resolved.

If the main reason IPs were not resolved is due to lost/corrupted logs or logs lost to normal file rotation, they have little to no bearing on accuracy. This is what people demanding shorter log retention are banking on and if TSI fails to get their costs paid by Voltage, they might decide to do just that so next time Voltage or other troll comes with 2+ months worth of IP logs starting weeks ago, TSI will be able to tell 'em they cannot do anything with logs older than 21 days and chuck the whole thing out instead of wasting $200k again.

JMJimmy

join:2008-07-23
Reviews:
·TekSavvy DSL
said by InvalidError:

said by JMJimmy:

they gave the court a major point to consider: Only ~50% of the IPs could be associated. That has huge legal significance because this is a civil case and the odds are important.

The reason(s) why 50% of IPs could not be resolved to an account are important too and may not necessarily have any negative impact on the accuracy of the other 50% that were resolved.

If the main reason IPs were not resolved is due to lost/corrupted logs or logs lost to normal file rotation, they have little to no bearing on accuracy. This is what people demanding shorter log retention are banking on and if TSI fails to get their costs paid by Voltage, they might decide to do just that so next time Voltage or other troll comes with 2+ months worth of IP logs starting weeks ago, TSI will be able to tell 'em they cannot do anything with logs older than 21 days and chuck the whole thing out instead of wasting $200k again.

Do we know this is the case for sure?

Even so, it does go to accuracy. If log files are being overwritten prematurely with new data due to rotation or files are being corrupted that points to issues in the storage/software/processes/etc which calls the entire data set into question.

InvalidError

join:2008-02-03
kudos:5
said by JMJimmy:

Even so, it does go to accuracy. If log files are being overwritten prematurely with new data due to rotation or files are being corrupted that points to issues in the storage/software/processes/etc which calls the entire data set into question.

If some log files are missing or corrupt to an extent that their content cannot be trusted or reliably interpreted, you simply chuck them out and that does not affect the integrity of remaining good files.

The holes this may create in individual sessions' histories where the start/end of a given session may get lost or becomes uncertain are likely responsible for a large chunk of the 50% that got flagged as unresolved for other reasons than logs simply having expired.

If TSI's process is to piece together the history of each session from start to end, dropping any IPs where that history is incomplete, the likelihood of mistakes should be practically zero. That would both explain why TSI's alleged cost per IP is so high and why so many IPs were flagged unresolved.

JMJimmy

join:2008-07-23
Reviews:
·TekSavvy DSL
said by InvalidError:

said by JMJimmy:

Even so, it does go to accuracy. If log files are being overwritten prematurely with new data due to rotation or files are being corrupted that points to issues in the storage/software/processes/etc which calls the entire data set into question.

If some log files are missing or corrupt to an extent that their content cannot be trusted or reliably interpreted, you simply chuck them out and that does not affect the integrity of remaining good files.

The holes this may create in individual sessions' histories where the start/end of a given session may get lost or becomes uncertain are likely responsible for a large chunk of the 50% that got flagged as unresolved for other reasons than logs simply having expired.

If TSI's process is to piece together the history of each session from start to end, dropping any IPs where that history is incomplete, the likelihood of mistakes should be practically zero. That would both explain why TSI's alleged cost per IP is so high and why so many IPs were flagged unresolved.

True but if start/end aren't being logged reliably, who's to say the the "good files" don't contain errors as well? It really depends on the specifics of why those 50% were unresolved.

A hypothetical: Log Y... User 1 starts session with X IP log Y isn't terminated properly, User 2 gets X IP and system continues in log Y and log ends properly this time.

I'm not saying that's what happened, I have no clue of the specifics, only TSI does. What the court knows (at this point) is that 50% of the IPs submitted can't be connected to a given account. That casts a lot of doubt on the voracity of Voltage's evidence. We don't know if there was 1 reason or dozens for the unresolved IPs.

InvalidError

join:2008-02-03
kudos:5
said by JMJimmy:

True but if start/end aren't being logged reliably, who's to say the the "good files" don't contain errors as well? It really depends on the specifics of why those 50% were unresolved.

A hypothetical: Log Y... User 1 starts session with X IP log Y isn't terminated properly, User 2 gets X IP and system continues in log Y and log ends properly this time.

In general, computers do not make "errors" and when programmer fudge something simple like a log file up, it usually leaves some rather obvious signs such as garbled text, weird text alignments, unexpected new lines, etc. that have a tendency to break scripts or trigger parsing errors/warnings to get the operator's attention.

As for your hypothetical case, the "worst" case is 'A' ends up in the "unresolved" since the session's end time is unknown, which would be perfectly fine. Nothing happens to 'B' due to A's infringement since the ISP already knows for sure that 'B' did not own the IP when 'A' did. (Or nothing happens to 'A' due to 'B''s infringement since the ISP knows 'A' certainly no longer owned the IP by the time 'B' got it.)

Considering the amount of mind-numbing work involved, it is to ISPs' and their subscribers' advantage that the decision threshold to abandon researching individual IPs be as low as good faith can reasonably allow. Any reasonable doubt about the current IP-timestamp and the lease's timeline? Mark as unresolved. Next.

JMJimmy

join:2008-07-23
Reviews:
·TekSavvy DSL
Ya - I'm a developer so I am vary familiar with these sorts of issues.

If the log system references an existing file (as you suggested) it could be appending to it instead of overwriting it in which case it might not be obvious. User A could be the one infringing and User B gets the notice... again this is all hypothetical.