Ran across this site whilst poking around:

[http://www.web1000.com/]

They offer free hosting, and evidently suffer greatly from child pornographers trying to get away with stuff.

This raises a related question. How does an operator of dedicated servers offering reseller accounts keep on top of the activities of the numerous and unknown customers of the resellers? A sleazy reseller may slip through, or a sleazy customer may slip by an ethical reseller, putting an entire server at risk of sudden hard drive confiscation by the F.B.I., or account cancellation by the data center for a spammer or other fraud artist.

Is there a fairly standard set of tools and procedures for protecting a business and its servers against the serious problems caused by outright criminals, or by otherwise sleazy customers and/or resellers?

Yeah a nice AUP!

Simply raising fire and brimstone in the AUP is enough? What about being able to keep up on gigabytes of material, ultimately sprawled across multiple servers? Perhaps a tool to scan all known hosted domains for giveaway words and phrases?

Is it not reasonable to be paranoid about spammers and other problems when one that slips through, even for a short time, can destroy a small startup business?

Originally posted by badbaker.com
Yeah a nice AUP!

Content management is easier than you think, I've written programs to do it before on a regular basis - if you're not a programmer you can do it with any search utility.

The program I made was pretty simple, written in Visual Basic 6. You create a text file of "naughty words" - words directly related to the type of content you don't allow on your servers. Tell the program your webspace starting directory (the top directory in which all customer sites are below) and it will go through every .html, .htm, .asp, .txt, .pl, .cgi, .php, etc. file in all sub-folders and flag any files that contain those words.

Then it looks up the owner of that site in the database (based on the directory) and builds a small profile which (at the end of the report) is forwarded to an e-mail address for review.

The report is an HTML email with check-boxes and submit buttons - the recipient of the message (me or my staff) can go through and manually look at the pages that were found containing the buzz words and verify whether the word is used appropriately or not. If it's considered "OK" the person can indicate so in the form and hit the button, and it will tell the program ignore that site in the future. Check the other box and it will send a message to the user notifying them they have illegal content and request that they remove it or get canceled.

Overall it's been a pretty reliable system. Of course you're going to get a few here and there that come up that aren't in violation (journal sites for example, where they may be talking about a particular subject that contains our buzz words). It's not like the words "adult" or "sex" or swear words are on the list because they're too common, but rather combinations that are used universally on adult sites, racism sites, things that are potential scams (I don't host anyone who is an affiliate of any MLM for example because too many of them are scams, and those MLM people are annoying anyway)

I run the program once a week on a random day. So far it's been working out well.

Ah! Most cool! Have read your excellent post carefully, and will do the same after research into possible pre-existing tools.

Agreed about the MLM people being simply too annoying, not to mention prone to harboring outright crooks. Will remember to politely ask in the agreements and in the pre-sales FAQ that they not patronise the business.

Originally posted by Indy4
Content management is easier than you think, I've written programs to do it before on a regular basis - if you're not a programmer you can do it with any search utility.

... [snip] ...

Then it looks up the owner of that site in the database (based on the directory) and builds a small profile which (at the end of the report) is forwarded to an e-mail address for review.

... [snip] ...

... potential scams (I don't host anyone who is an affiliate of any MLM for example because too many of them are scams, and those MLM people are annoying anyway)

I run the program once a week on a random day. So far it's been working out well.