Battling Spam and the Ring of Gyges

Commentary Rant

Humanized is currently fighting a battle against spam on the comments section of this weblog. Automated spambots have been posting hundreds of “comments” a day, which are content-free posts under fake names containing links to dubious merchandise. They are often obscene and sometimes more offensive than dead-baby jokes.

There are four strategies we could use to keep spam comments off of our page.

Method 1: Automatic filtering. Using various mechanical means (keyword filtering, IP tracking, etc.), we automatically classify each new comment as “spam” or “not spam”. This method has the advantage of introducing no interface overhead to the user. Unfortunately, it’s difficult to achieve. Like biological organisms, comment-spam messages have evolved camouflage in order to hide from their natural predator, the spam filter. Most comment-spam messages are mundane sentences like “Hey great blog, take a look at this link I found”, which is identical to a legitimate comment. Only the commenter’s name and the destination of the link give clues to its spammitude. This makes it very hard to design an automatic filter which will not be plagued with false positives and false negatives.

Automatic filtering is the first method we attempted to use for our site. However, as traffic increased our filters were soon overwhelmed. We were spending hours a day manually deleting all the filth that got through the filters. We decided to look for a better option.

Method 2: “Captcha”. This oddly-spelled word refers to those systems requiring a user to read characters or a word from a distorted or fuzzy image, and re-enter them. Like automatic filtering, this is another attempt to reduce computer-generated spam. The theory is that we can prevent automated posting of comments by requiring users to perform a meaningless task which is easy for a human being, but impossible for a computer program. This theory assumes that posting spam manually, one-by-one, is not worth a spammer’s time, so he will give up, while legitimate commenters will be willing and able to decipher the captcha.

However, this might not be a good assumption. A bedrock principle of humane interface design is that computers should not force users to perform arbitrary and irrelevant tasks; if a user’s goal is to enter a comment, it’s hard to think of anything more arbitrary and less relevant than “Please decipher these squiggly numbers to prove you are not a computer.” Some people will get annoyed enough by this requirement that they will give up on commenting. Other people — the visually impaired, for instance, who use screen readers — will find the requirement impossible. Since humane interfaces are our first priority, we decided that the captcha is unacceptable for our site.

Method 3: Moderator approval. After a user submits a comment, but before it appears on the site, one of our employees must read it and decide whether or not it’s spam. Only approved comments get posted. This makes extra work for us, and it introduces a delay before the comment appears, but it is the only method which keeps out all spam, doesn’t block out legitimate messages, and does not force the user to jump through any interface hoops.

Method 4: The accountability system. Make our best attempt to hold commenters to fixed identities so that we could hold each one accountable, and kick off spammers. From an interface standpoint, this would require making the user jump through hoops (i.e. a login screen) before posting. But it gets worse: To discourage spammers from simply using disposable accounts, we would have to use a multi-step account-creation process, perhaps involving email-address verification, in the hopes that a real user would have the patience to go through the process but a spammer would not.

This was unacceptable to us; we want to encourage comments by making them as easy as possible. If users have to go through a multi-step account-creation process with email-address verification before they can post their first comment, they are likely to decide that it’s not worth the trouble! We expect that our average user already has account names and passwords for dozens of other websites and services, and isn’t particularly interested in creating yet another set just to be able to comment on this one blog.

Mutual Accountability and Reputation

The accountability system is especially interesting because it is a fundamentally different strategy from the others. Can you see why? The other three methods have one thing in common: they assume that users are untrustworthy. A central authority — either a human moderator, or some sort of algorithm — is used to forcibly restrict what users can and can’t do.

But accountability systems are different. They assume that most users are trustworthy and should be free to do whatever they want. When a user abuses the system, by posting spam for instance, other users can choose not to interact with the abuser (by blacklisting or “killfile”-ing him.)

My favorite example of an effective accountability system is eBay’s feedback system. If someone rips other people off, he will accrue negative feedback and no-one will want to do business with him. eBay doesn’t kick people off, or punish them in any way, for having negative feedback: they don’t need to. The visibility of negative feedback to other users is enough. It’s kind of a “scarlet letter”. Allow users to identify the “bad guys”, and the users will regulate themselves. People on eBay take negative feedback very seriously.

Whether it’s called “feedback”, “reputation”, or “karma”, accountability systems work only when a user has a fixed identity for his actions to attatch to. If you can throw away an identity and start over every time you get bad karma, then the accountability system is significantly weakened.

I call this the “ring-of-Gyges effect”. Gyges was a character in a parable by Plato who found a magic ring that let him turn invisible. Since he could avoid all consequences of his actions, there was no reason for him to behave morally. Starting a new account is the Internet equivalent of turning yourself invisible to escape punishment.

This is why accounts with no positive feedback are regarded with great suspicion on eBay: because they might be a new name for an old user with bad feedback. This creates a bit of a problem for legitimate new users trying to get started on eBay; they may have to make several purchases first before anyone will trust them enough to buy from them.

Contrast eBay with a protocol designed in an earlier, more innocent era, and which therefore has no built-in accountability system: email! Email spam is a direct result of the ring-of-Gyges effect. If email accounts could be held accountable (no pun intended), it would be very easy to simply blacklist spammers. Users could choose not to interact with email addresses which have proven themselves to be badly-behaved, just as users choose not to interact with negatively-feedbacked eBay accounts.

But alas, email addresses are easily disposable, blacklists are useless, and spam is everywhere. Email’s failure is its lack of accountability. Even if we fix the holes in SMTP, we still have to deal with the ring-of-Gyges effect.

There’s a great example of how an accountability system can evolve over time. Wikipedia was originally based on the idea that anyone could edit anonymously. This worked well for a while, but because of recent increases in vandalism, Wikipedia has had to modify its policy slightly: now only registered users can create pages, and some controversial or recently edited pages get locked with this message:

Because of recent vandalism or other disruption, editing of this article by unregistered or newly registered users is currently disabled. Such users may discuss changes, request unprotection, or create an account.

Note the phrase unregistered or newly registered users. This sounds like a response to a ring-of-Gyges problem: newly registered users might be old vandals with new names. You must establish yourself as a trustworthy member of the community before you can edit the locked pages.

Accountability and Interface Design

Holding users accountable always requires authenticating their identity, which nearly always requires them to go through extra steps before they can begin using a system. Thus, authentication incurs an interface cost. Plus there is an additional interface cost to create an account in the first place, as I mentioned earlier. Not only do users have to learn and navigate the account-creation interface, they must also come up with a new username and password (or recycle an old one, which we all know we’re not supposed to do, but we all do anyway…) This cost is generally enough to push some users away from participating in a site. When the whole purpose of the site was to encourage user interaction, this makes the developers very sad.

(You might be able to think of some exceptions to the authentication-requires-extra-steps rule. For instance, remote login using ssh with a public/private keypair and a stored passphrase. A user set up with a system like that doesn’t have to enter anything to login; however, the process to initially set it up is extremely archaic and inhumane, and not at all appropriate for simple things like commenting on websites!)

If only there were a concept of a universal online identity, accepted everywhere, and defended rigorously! Then users could authenticate once per browser session, and be recognized at every site they visit. For users, the annoyance of creating passwords and logging in to every site would disappear. For webmasters, the annoyance of implementing a custom user database would disappear. Interfaces would be simplified. The reputation of a user, attached to this identity, could extend across all sites. Abusive users could be identified, blacklists could be shared, and users could choose whether or not to communicate with blacklisted entities.

(Maybe there is an alternate universe where AOL invested in broadband technology early and then took over the Internet. In that universe, a user’s AOL screen-name is a universal identity recognized everywhere, and AOL can be stingy about giving out new names, to prevent the ring-of-Gyges effect. But this alternate-universe Internet would also have drawbacks compared to ours: lack of privacy, and the potential for abuse of power by the all-powerful gatekeeper AOL. Would it be worth it? An interesting scenario for you to ponder.)

There have been attempts at creating such a system. TypeKey is one example. Bloggers can use TypeKey authentication instead of implementing their own. Not only do they save development time, but users with TypeKey identities don’t need to keep creating new passwords for each blog they want to comment on. This is a good idea which faces a classic chicken-and-the-egg problem in getting people to adopt it. But if TypeKey or some equivalent does become widely adopted, it might help sites like ours to provide accountability with a minimum of interface cost to our users.

What Next?

The evolution of accountability systems seems to me to be key to the further development of social networking. There are several trade-offs involved. Internet users love anonymity. But they hate spam, trolls, wiki-vandalism, and people ripping them off on eBay. They also tend to dislike having a central authority arbitrarily tell them what they can and can’t do. I’m generalizing a lot here, but I think most Internet users prefer democratic, decentralized solutions wherever possible.

When users can work together, freely sharing knowledge, without centralized authoritarian control, they can create great things, like Wikipedia, using their collective intelligence. But when users aren’t or can’t be held individually accountable for their actions, bad behavior goes unpunished, and a thriving community may degenerate into the Internet equivalent of an angry mob. A central authority can prevent the angry mob effect, but it can also stifle the creativity of the collective intelligence. Mutual accountability can eliminate the need for a dictator, but sacrifices anonymity.

The ideas of accountability and identity, and the technological strategies to support them, are changing rapidly, as are the trade-offs involved. This article will be obsolete by the time you read it. As you travel around the Internet, look at the different strategies that your favorite social sites use to prevent abuses. Do they use a moderator/central authority? Do they use an automated system like captchas or filtering in an attempt to mechanically prevent abuses? Or do they require you to create an identity which can accumulate positive and negative reputation? How well does it work, and what does it do about the ring-of-Gyges effect? And what sort of user-interfaces do your favorite sites use to minimize the cost to you?

For now, Humanized has decided that an accountability system is not the way to go for our weblog; we are instead using manual moderation of all comments until such time as we can come up with something better. Manual moderation of comments means setting ourselves up as dictators of what can and can’t be said. This has its own problems, but at least it will allow us to eliminate spam without making our commenting interface more difficult for users.

You forgot one of the biggest “universal identity” solutions out there - Microsoft Passport.

I think we already have a good basic key for universal identifitication, and you glossed over it - PGP or GPG public keys. The problem (and the reason you glossed over it) is that there doesn’t exist a humane interface for dealing with those keys. But if people were to make tools that users could actually (securely) use, they would be a marvelous solution to the identity problem.

Here’s an idea for a solution. It’s very much a partial solution, however. The key to preventing spam is identity, but there are two problems to identity online - the first is that creating and maintaining an identity differs from site to site, making it tedious and difficult for “actual” users, and the second is what you have called the ring-of-Gyges problem: the fact that you cannot differentiate between a “bad” user and a new user. This idea only solves the first problem.

Instead of having a universal identity, web browsers already come with the functionality to transparently (and therefore humanely) keep track of a user’s identity - cookies. You can give a user a cookie that will essentially act as that user’s account and login. All a user has to do to receive a cookie is leave a comment!. This way you can cut your manual moderation down by simply whitelisting all users who have an accepted cookie.

Another idea, as reported on Slashdot, is to add (perhaps randomly) fields which are not shown to the user’s browser, and which will cause submissions to be rejected if they are filled in. There are two problems with this: it is an “arms-race” style solution that can be beaten, and it may render the page incorrectly on older browsers than don’t handle CSS. However, it does protect you from web-crawling spambots (at least until those robots become sophisticated enough), and it seems that “humane” interfaces are not too concerned with being usable on very old technology.

Why not take a 2-pronged approach- introduce an optional system that adds functionality such as coComment - http://www.cocomment.com/ and keep the manual moderation in place for people that aren’t willing to sign up.

You get the best of both worlds and reduce the amount of manual labor required. Also, with coComment, users get the added benefit of being able to track all their comments across the web in one place. They may not offer the level of accountability that you would like, but it would be worth checking out.

Of course, this could also be applied to any of the other methods — make them optional (with benefit, i.e. faster posting, comment tracking, etc.) and manually sort through those who don’t want to participate.

That’s an interesting point on the distinction between the two types of trust mechanisms. Oh for a world with a single login — but without the resulting privacy and security concerns.

Anyway, one suggestion for your problem: Akismet (http://akismet.com) provides an API for comment filtering. It analyses comments from millions of blogs, and so can very quickly recognise new bots after the first few blogs tag them as spammers — so you get the benefits of an automated filter, without the inaccuracy. Unfortunately you may need to write your own interfaces if you don’t use one of the “standard” blogging systems, but once that’s done I think you’ll find the spam problem essentially solved.

I can’t tell you how much I sympathize with you guys on this. I encountered the exact same dilemma with the raskincenter.org forums. I came to the exact same conclusion: I must temporarily moderate until something better comes along. The only difference is that I screen user accounts and not individual comments. (Well, since I’m not getting paid I slipped further and am making new–real–users email me in order to get activated.) There are better options that would effectively solve the problem relatively nicely, but it requires a lot of work and risk–I have neither the time nor authority to mess with it now :(

But I think this problem is bigger than anyone individually. This is a fundamental design flaw that will need to be addressed in the next incarnation of the net (rumors of a “3.0″ are already making rounds, something like a public Internet2).

I have my own ideas about how this should be solved, but who doesn’t. All I know is that It’s going to be really interesting to watch how it will be really solved… and I think IPv6 will have a lot to do with it–it makes completely unique identities possible (here’s a link to the rest of my idea :)

Interesting problem indeed.

as a commenter, I would rather decipher fuzzy letters and have an immediate posting, than to have a delayed posting where you have to go through and check the human-ness.

in your situation it is probably best to do what you have been doing because immediacy of a posting isn’t that critical, yet. and requiring someone to need to sign up is a hassle.

the most ideal situation would be for the operating system to have a built in ‘identity’ in it that would be something to be used all over the web. if you check out the website http://kevo.com, they have a pretty interesting karma system, it’d be interesting to study how this might be possible from world to word, rather website to website. sorta like microsoft’s passport or something? but open-source… ? that comes with all sorts of baggage though… ?

Some of the advantages of ‘creating’ a user are that I don’t have to fill out that form more than once, unlike this system, where I need to fill in my name, email, and url each time, it would be nice if it remembered me, that would be humane.

Your system is optimized for users who have just discovered the site. Which is great because you’re more likely to get comments from people from people not willing to hassle with troublesome interfaces.

Another thought: creating a user account can be as easy as filling in the information that I had to fill out in this comment form, it doesn’t need to be lengthy. I don’t even need a user name, my user name can be my full name, it could even create a user account for me… just like when I introduce myself to someone I don’t have to tell them to remember me, they just do! unless i’m not interesting, and then they forget me, but that is usually only after a while, and they will still likely remember my face, at least.

this still doesn’t solve the problem of spam… the best way probably might be to just let the spam through, and then let everyone democratically vote to remove comments, or ‘hide them’, or put them in the comment “trash can” :) and you don’t have to sign up to vote either… wonder how this would work… and what kinds of problems that might introduce…

Anyways, to humane interfacing! :)

Hide a form field from real users labeled something that a bot would fill in. If filled in it’s a bot.

http://isc.sans.org/diary.php?storyid=1836

Other than an AI, I’m not sure how one can achieve your spam-elimination goals without a reputation system or Turing test system. I agree that any sort of reputation system sacrifices some amount of privacy; I’m pretty sure this is a theoretical requirement. I think you are overlooking the value of incomplete reputation data.

Let’s say you push a cookie out to clients that uniquely identifies them; you set it to expire in a week, month, year, whatever. Every time you get a valid comment by said person, you increment some counter in a database that corresponds to the user’s ID. At some threshold, you skip the moderation step. This only works for regulars, so it could be quite valuable or quite useless to you, depending on your traffic. If the cookie disappears, nothing much is lost. The system is transparent to the user.

The next step up is to allow people with a certain threshold to moderate, if they so choose. Perhaps you monitor their moderations until they’ve proved they are good moderators, by another threshold. You could include a CAPTCHA-based conflict resolution feature, to weed out the moderators who abuse their power.

I would suggest suggest adding an *optional* sign up. Comments from these people would not need moderation, and they could even be tasked with helping out with moderation of others posts.

Akismet, guys. It’s awesome. It’s not just for WordPress anymore. Minimal moderation is required on the blogger’s part, and the commenter doesn’t have to do anything. Works like a charm.

Thanks again for your insightful and interesting ruminations.

A few comments in return.

CAPTCHA does not necessarily have to be hard to read. Using a simple image of 3 to 4 numbers on a plain background will stop all comments from current spam bots, or so I am guessing. There is a huge cost in deciphering (OCRing) images, no matter how simple they are, and I’m guessing most spam bots would not bother with it, especially if it is only for comment systems. Where they would try would be on services such as Gmail, where they can automatically get accounts (with manual setup of a bot to specifically target that web service), and use that to spam people who would not normally receive such and thus pay attention.

CAPTCHA is not so much a pain when you have to do it once on sign up for a service. But you’re absolutely right that it is annoying to have to do it on a “simple” comment.

Something else you could try would be to moderate all comments not posted by a previously posted “Name” / “Email Address” pair. As you do not post email addresses to the site along with comments this would mean that only first time posters, and spam bots, would need to be moderated. Not quite perfect but it would help, and make your users not feel they are being moderated.

Maybe you could check to see if people have a “Gravatar” and give that post a highlight on your system to make it more likely that it is not fake.

You could make it even easier for people to not have to be moderated by, for those who have JS enabled, send an extra “JAP-T-SAP” (umm - JavaScript Automated Public Turing test to tell Spammers apart from Public - sorry), almost the same way as CAPTCHA (store server-side a generated code for each page request, then when the form is submitted call a bit of annoyingly complex JS that recreats and sends the code, the JS being created when the page was served).

Now, the single-signon-services.

As far as I am aware this is what Microsoft was trying to do with their Passport system (always based in Hotmail), which would have worked well if they had not restricted it’s use so much, even letting users log into their computer systems (yes, Win XP only).

OpenID (http://openid.net/) is a service that can bee used for this, and seems like a good one, and is on use on a few services such as LiveJournal (I think the creator was from there).

What really intrigues me is I-names (http://en.wikipedia.org/wiki/Iname), which I do not “trust” as yet as I have not heard anything about them on blogs / media. But if they do work, then they would let you have single-signon as well as email routing etc with a username such as “=tarwin”. If anyone knows what’s really happening with this I’d be really happy to hear your thoughts.

As an interesting side note, when Firefox came out with ‘inline spell checking’ I was dubious as to it’s usefulness, but have since using it found it forcing me to type comments such as these without the obligatory “teh”s and such which is hopefully making what I say a whole lot more readable (except for those stupidly long sentences).

Last thing; please let me resize this text box, or at least make it’s size a percent or em, it’s late and I want to be able to see the text I’m writing through the blur, but not one letter at a time.

Thanks. Thanks for the thought provoking.

After posting my last comment I was quite horrified.

I was send to an unstyled, unfinished page with a “thank you” response.

I would much prefer to enter a simple CAPTCHA than be “subjected” to that. Which brought a new idea to mind.

Let users enter a CAPTCHA if they want, this gets them past moderation. If they don’t then they just have to wait in line with the spam, but do still get on. Let the users decide. How you make it clear that they can enter the CAPTCHA but don’t have to without putting in a whole lot of extra crap is beyond me right now.

And when one thinks about it, if someone is going to make a “worth-while” comment, something that isn’t just “:( too bad” or “yay!1!!1!” then it might be worthwhile to remove this noise by forcing a CAPTCHA. It doesn’t hurt to enter an easily read 4 letters before you post, if your making a “real” comment.

I realise those things in quotes are probably something that will create a lot of argument.

The fact that I cannot read other peoples comments before you “moderate” them means I’m speaking to myself, until the point when they all suddenly appear, all at once.

PS: Do you realise that as you type a lot the text goes off the edge of your design on the right hand side (Live comment preview)?

Hi Jono,
What a great breakdown of the various options for dealing with spam management on a blog (or similar application).

Regarding all this, I understand that “Humanized” is about the humane interface and puts the user first and as a starting guideline that completely makes sense.

As someone who has been blogging a bit and has dealt with the spam problem head on, I too started out (and actually still do) moderate comments on my personal blog (synapticburn.com) and on IxDA’s Resource Library (resources.ixda.org).

In both instances just using moderation became such a burden to the human beings behind the scenes that two options were put in place.

Fro my personal site I decided to use CAPTCHA. It reduced my spam to a trickle, but some still gets through, which is why I still manually moderate.

For the Resource Library, we were getting so much spam and so little participation at the time that we decided to shut off commenting completely. When I say a lot of spam, I’m talking hundreds of spam messages a day.

I don’t wish this on you, but when designing a system, you need to realize that there is more than one stakeholder of your system.

How much resource do you have to dedicate to manual moderation of this blog?

And, you didn’t account for the problem that people think of blog posting as an instant action, so how much expected delay do you expect people to tolerate as part of the pain of waiting for the moderator.

Another place where I have this experience is on the IxDA mailing list. We get like no spam, but the concern of spam is so high that I set it up so that first time posters get moderated. After that, its on an accountability and monitoring basis. But b/c of the moderation I often have to deal with duplicate postings because people don’t want to wait.

Just wanted to point out that there are more human beings to the equation and consequences for the moderation approach that you might have thought of as a team, but didn’t include in your article.

– dave
(waiting patiently to see my comment)

I think you may overestimate the “inhumaneness” of CAPTCHA. Personally, I always like that part of filling out a form. First, I get to prove I’m a human, which gives me a certain amount of speciesist pride. And furthermore, after filling out this form that has required me to make all kinds of choices — should I use my full name? My main email address? What will I comment? — when I get to the CAPTCHA question I go “finally, something I know the answer to!”

A relevant quote from a design column you and I both read: “The lesson here is that you have to be very careful each and every time you ask something of your players. Make sure that there is a high correlation between how much work you ask of them and how much they get out of it. When the ratio gets too high, you have to reexamine why you?re using that mechanic.”

Okay, so a more elaborated post.
Firstly, the current method discourages conversation. I have no idea what the past 10 comments are that mine may follow, and so I am not inclined to respond to someone as the thread of conversation may change.
Secondly, I think it is interesting to look at codinghorror’s captcha. It is one word, very clearly typed and constant. However, it has proved very effective.
Re: Matthews suggestion; http://www.codinghorror.com/blog/archives/000715.html has some interesting comments on Akismet style systems, as well as other spam prevention methods.

ICR, great point about the lack of fluidity in the comment area. I came back a few days later, and was expecting my comment to be posted at the top, b/c the comments are was blank when I posted it. But I found I was closer to the bottom. I almost didn’t even bother looking for it.

And then your comment about how to have a conversation when there are 5 people all saying something at the same time and you can’t hear them until 10 min. later (or really a few days later).

I think this brings to mind the whole issue of humanizing the tree without humanizing the forest at the same time.

From my personal experience as a blogger and a commenter, I’ve made a lot more mistakes trying to complete CAPTCHA forms whilst trying to comment than Akismet has made deciding whether comments are spam or not.

And I agree with [IRC] above that moderation means that you don’t really know where in the conversation your comment will fit… which is off putting.

I’ll be resisting using CAPTCHA as long as possible and Akismet does an almost flawless job for me in the meanwhile.

I think the solution here is actually technology and not so much the interface… except for keeping the interface out of the way of the user as much as possible.

While I like the core of your post, I don’t agree with your positive view on eBay’s (and similar) rating system.

I’ve just checked back with ebay.com to see whether the situation there is like on ebay.de, and sadly enough, it is.

There is not any actual feedback. Stereotypes abound, like “great seller!”, “excellent ebayer. nice to do business with. A++++”, “A+++ Thanks again!”, and so on.

Why? As soon as seller and buyer are both ebay users, there is a double-bind that forces them to hype everything.

It’s like nuclear deterrence: don’t bomb my sales, then I won’t bomb your credit.

As an suffix to my earlier comment: I’ve found someone doing exactly what I was proposing for the Javascript blocker. It’s here: http://15daysofjquery.com/examples/contact-forms/

What could be used to make this completely non-JS required is to have a CAPTCHA that is removed with JS, so when JS is disabled it just shows CAPTCHA.

Should you be spending all of your time moderating comments to make adding comments more Humane or should you be spending time working on Enzo and other Humanized products to bring about a Humane world?

All problems should be dealt with strategically, rarely can one kill the opponent or problem, just maximize your territory. Right now they are winning in terms of your time and degrading the quality of discussion we can have.

1) Use transparent spam control methods to cut spam down to a minimum.

2) To stay compliant with Jef’s law, “Any system shall not harm your content or, through inaction, allow your content to come to harm,” have a simple CAPTCHA offered to positive spam results which would also place that user on a white list.

Depending on the accuracy you may wish to layer different methods of automated spam detection, make the white list still have some basic protections, and introduce community flagging. You may even drop the remaining queue on some volunteer admins from here or the Archy project.

-ZL

Great post about the most salient lacking building block of the internet : identity.

I think a good identity should have three properties :
Be Owned by the user
Be Hard to falsify
Be Resolvable to your universe

Thus for me the solution is personal web domain aka you.something.com
You should own the domain forever
The domain should be backed by a digital certificate
The domain would be highly resolvable via HTTP to your private digital life.

Phone Number is not owned
Email is not resolvable except for messaging
Hail to Private Domain for personal ID !

Thanks for putting all this work into the forums.

Why not keep the forum the same as it is with the only differences being you have to register your email address before you can post a message. Since you are already asked for an email address there will be none of that logging in and out malarkey. The only extra thing I would suggest is a check box that allows your email to be visible to the forum. Since there are a lot of bots out there that harvest email addresses I think this should be set not to show your email address by default. That?s why I never give a valid email address when posting, I get enough Spam as it is.

What do you think?

Silly me I just noticed that email addresses do not appear to be displayed. I had presumed that when the persons name was a link that clicking on it would allow you to send them an email but I guess that must be for their URL. Would it be more humane to show peoples URL?s as links with the word website: beside it?

If Heuristics or a transparent CAPTCHA work 99% of the time then for the vast majority we are actually degrading the experience since only 1% will ever run into a CAPTCHA.  Who is really “winning” in that case?  I know this is a rant,  but it seems too focused on total annihilation of the problem instead working towards a pragmatic goal.

 My earlier proposed solution still presents a usability problem since 1% of the users would run into a serious, high error rate, data losing mode.  Having habituated another path they are very likely to completely miss the resulting CAPTCHA page, “failing” the test.  Instead, use JavaScript to “obscure” the Post button if the user “fails” the invisible form field, Bad Behavior, or, possibly, any tests a heuristics engine would run on a post, or parts of it, when streamed via XMLHttpRequest- links being an example.  Although, the last option does have several drawbacks and may cause further errors.  For computers with Javascript turned off the CAPTCHA would be the default.

Submitted posts could then be tested by a heuristics spam engine, Akismet and Spam Karma being an example.  If a comment fails post-submission testing a combination of things could be used to protect against false positives.  One is to have a large notice offering a CAPTCHA tagged to that user on top of the page and/or as a place holder for their comment, emailing them, community moderation, white lists, gray lists, etc.  What to use will always be a balance between mucking up the comments section and loosing user data and should be determined by the effectiveness of real world testing.

A discussion on a humane CAPTCHA would be useful.  One idea is to use simplistic pictures or illustrations, asking the user to identify the picture or something about it’s contents.  The same field could also accept answers to a “hidden” disabled version.  It is also a probable long term solution since lots of big companies have largely failed at having computers identify objects .

By implementing these spam protections (with quite a bit of tuning) you could have perfect compliance with 99% of users while not directly violating any major usability laws for the other 1%.  I think that would be a winning score.

BTW, anyone else here play Go?

Looks like you have fixed the problems. No more delays in postings! Thanks.

I would really appreciate it if you could tell us what you did to fix your problem?
I’ve added this post with a different (though valid) email address, so I’ll be interested to see if it turns up straight away.

Tarwin: After considering the many excellent ideas suggested in this thread, we found a very simple heuristic which can rapidly categorize the majority of posts.

Ever seen a spam comment without a link in it? There would be no point. Therefore, we can assume that posts with no links are legitimate, and get posted immediately. Posts with links get sent to an automatic filter; it is quite good at correctly flagging the majority of spam, and only the posts which it considers marginal get passed to a human. This drastically reduces the workload on the human moderator. It’s a multi-step, composite solution.

Also, I’d like to respond to Rolf’s comment about the uselessness of eBay feedback comments. He is quite correct: most of them are completely useless and inflated fluff of the “A+++” variety. However, the important thing is not the content of the messages, it’s simply the number of positive and negative comments. If an eBay user has 1,000 positive comments and no negative, I don’t need to read the contents of any of those comments to be able to judge that user trustworthy. That’s what I meant when I praised eBay’s feedback system.

Well, how’s that for comedic timing, folks? I say there’s no such thing as spam without a link in it, and then what appears immediately following my comment? Bah humbug!

That’s a great idea, I really like it. As you said though, there is a real problem when more people start using this heuristic as a default as people will start to spam with things such as wwwdotmyspacedotcom. I realise it’s hard to read but so is most of the spam that I thankfully don’t get in my email anymore.

Just a thought.

“Reduce WordPress Spam By Creating Fake Comment Forms For Spam Bots” http://www.onetipaday.com/2007/02/16/reduce-wordpress-spam-by-creating-fake-comment-forms-for-spam-bots/

This certainly makes a good point. I’d like to see the response of others on this topic. Makes interesting reading.
free software

rosie@triad29.com

rosieponder@verizon.net

Not only do they try to rip you off, they send your email out and you get a ton of junk mail.