May 26, 2005

The Long Tail of Tags, or, Spamming del.icio.us

Recently I noticed a whole bunch of links to a company called Airset when searching for del.icio.us bookmarks with the "evdb" tag.

Why would EVDB be associated with Airset, I wondered?

Strangely, all these bookmarks had the exact same number of contributors . . . 23:

And not only had exactly "22 other people" added the same entries in del.icio.us, but many had used the same tags: "online web calendar Treo Sunbird EVDB Blackberry mobile phone cellular".

Most of these users, like "pilotdad3", "jimdefrisco", "paqstuff", and "edelsesto", have only added one bookmark to del.icio.us, namely, the link to Airset, along with its long tail of tags.

One of these "22 other people" is pbhurley, who, if you Google for "hurley airset", you learn is Patrick B Hurley who works at, yes, you guessed it . . . Airset. In fact, he's their media contact according to Airset's own site.

Hurley, who also has only contributed one bookmark to del.icio.us (at least under the pbhurley user ID), takes the cake for the ultimate Airset tag spam:

productivity, software, tech, technology, tools, todo, free, asp, internet, 43Folders, online, web, calendar, Treo, Sunbird, EVDB, Blackberry, collaborate, collaborative,web-based, mobile phone, mobilephone, cellular cellularphone, cellphone,cell phone,PIM, address book, addressbook,contact list, to-do GTD, Getting Things Done, calendars, airena, smartphone, PDA, beta, Palm, Outlook, 43 Folders, webapp, app, shared connected , calendar,tiddlywiki, social networking, organizer, lifehack

Some of the other userIDs appear to be Airset employees as well: "dfults" I suspect is Doug Fults; "edelsesto" I suspect is Eric Del Sesto. I'm also gonna guess that "BrianDoc" is Brian Dougherty, founder of Airset.

Question is: will all this tag spam help or hurt Airset. We'll see.

Posted by brian at May 26, 2005 08:10 AM

Comments

Did you email Joshua about this?

Posted by: Ted at May 26, 2005 02:08 PM

nah. I figure Joshua has more important things to worry about, and karma will take care of things :-)

Posted by: Brian at May 26, 2005 02:42 PM

Since I'm mentioned in the post above, I thought I would respond. I would have preferred to have been contacted personally to respond to the author's questions / assumptions but it appears too late for that. Next time, give us a call. I know you have Brian Dougherty's email address if not his phone number.

It is true that several AirSet employees -- myself included -- recently opened delicious accounts and slugged in an AirSet bookmark but not for the motives ascribed to me / us. We would never consciously do anything that would be inferred as spam or gaming the system. If we breached etiquette, we did so naively and will own up to that. If we need to remove all the entries to demonstrate our integrity, we'll definitely do that without hesitation.

I just opened up a delicious account very recently
which is why I only have one entry. I certainly plan to add more when I have time to test it out and see if I like it. I felt nothing wrong with designating AirSet as my first bookmark -- kind of like typing your own name into Google on your first visit and seeing the results -- and tried to assign relevant tags.

I think something like EVDB is definitely a relevant tag as it is in the public event calendaring space and we're in the personal event calendaring space. A person interested in one could easily conceivably be interested in the other as complementary services. I think my tags are relevant or I wouldn't have included them. It's not as if I included a tag for Xbox or Podcast or something else that are very topical but not in any way relevant to AirSet.

I encouraged some of my colleagues to open delicious accounts to get a sense for another beta collaboration service (some already had accounts) and to slug AirSet in as one of their bookmarks (looks like in their haste that almost all of them used nearly the exact same tags). Also, we only have about 10 or 11 employees and I don't think they all opened delicious accounts so I can't explain why there are 22 people who referenced AirSet. I'll look at the full list of names when I have a second.

We certainly had no idea that our actions would skew the results as they have. As big as we perceive delicious to be, I wouldn't have even thought it possible.

Again, I wish you would have just emailed us or called us rather than unnecessarily assuming the worst in us. Next time you have a question, I encourage you to just contact us and get the skinnny first-hand.


Posted by: Patrick Hurley at May 26, 2005 02:52 PM

Gimme a break. Your lame excuse may pass for one person but all 23 of you? Hardly.

Posted by: dobbs at May 26, 2005 03:03 PM

I just went and checked delicious. By my count about 7 of the 22 AirSet bookmarks tracked back to anybody at AirSet. I recognize two other names as beta testers and they had bookmarked AirSet before we did. Don't recognize any others. They likely bookmarked AirSet in the last few days after we posted our bookmarks. And now.... I've said my peace. I'm going to get back to work.

Posted by: Patrick Hurley at May 26, 2005 03:30 PM

Wow!

You can choose to believe us or not but the simple truth is that we did not intend to do anything misleading or underhanded. We are a group of programmers who have largely been working in other areas (mainly mobile phone software) and not the web services area. We had heard about del.icio.us, checked it out to better understand the tagging taxonomy and Patrick suggested we all enter an AirSet book mark so other users who were looking for a web/mobile PIM might find us.

As Brian Dear can confirm, we've spoken on the phone about partnering with EVDB. We are mainly looking to network people's mobile phone PIM to everyone they care about and we view EVDB as a public event database we would like to let our users peruse to add events to their personal schedule. I still hope we end up partners. I thought adding evdb to our tag list benefited both companies, assuming someone who found us would also find them and vice a versa. Again, I view that we are largely in complementary spaces.

It was not until today that I realized a del.icio.us/tag/evdb lookup caused airset references to pop up all over the first page. I suppose if I were in Brian D's shoes and saw this, I too might have been a little pissed and jumped to the wrong conclusion. But honestly, this is more a sin of naivety than malice. I've gone in and removed the EVDB tag from my AirSet link and will instruct others in the company to do the same. I don't know who all the 22 people are since we only have 12 employees and I doubt more the 8 or 9 actually got around to creating a link on del.icio.us; but if people here got overzealous and used work and home accounts I will make sure we fix all of those as well.

Brian

Posted by: Brian Dougherty at May 26, 2005 04:04 PM

I figured it wasn't malicious, just naivete as you point out.

It's just that when anything like this happens more than a few times with a bookmark, and the end result is a sort of "monopolizing" of results like what's shown in the screen shot in the original blog post, it can really be misleading and frustrating for users. Chalk it up to a lesson learned about tagging.

Also, a comment on Patrick Hurley's statement, "I think something like EVDB is definitely a relevant tag as it is in the public event calendaring space and we're in the personal event calendaring space. A person interested in one could easily conceivably be interested in the other as complementary services." My comment is that it would not be accurate to state that "[EVDB] is in the public event calendaring space". We've never said we're limiting EVDB's scope to just public events. And event is an event is an event, and whether it's public, private, semi-private, semi-public is merely an attribute of an event. EVDB is for all events of all types everywhere.

Posted by: Brian Dear at May 26, 2005 04:49 PM

"I thought adding evdb to our tag list benefited both companies"

del.icio.us isn't about benefiting companies

Posted by: marcus at May 26, 2005 06:33 PM

I'm going to buck the trend here and say what the AirSet folks did was neither malicious nor naive nor spam -- there's absolutely nothing wrong with what they did.

del.icio.us leaves it up to its users to define the meaning of a tag. If I look at evdb-specific pages I see upcoming.org tags, if I look at trumba-tagged pages I see the EVDB homepage. The AirSet people believed that tagging with the name of similar services was valid -- this is clearly an interpretation that other users of del.icio.us share.

Whether the service is del.icio.us or Flickr, you are going to see many items that you disagree with in the shared-view, but your contrary opinion does not mean that the person(s) you disagree with are in the wrong. Flickr recognizes this by allowing the community to create groups that have stricter submission rules. Delicious has no real solution for this, which is delicious', not AirSet's problem.

In a week or so, most of the AirSet links will be on the second page, and hopefully by then the AirSet employees will be enjoying the Delicious service without having their tagging discipline publicly flogged.

Posted by: kwc at May 27, 2005 12:46 AM

I think we'll see more instances where lots of people tag the same link with the same tags. With the experimantal post to del.icio.us that is found somewhere on del.icio.us (I don't remember now where it is), you get at list of all your tags, and all the tags people so far have tagged a link with. So when you are number 45 linking the page, there is a pattern in keywords developing already, and most of the time I find that the keywords that are popular with the 44 prevoius taggers, are fine with me too, so I click them and add one or two myself.

Try for yourself with this page (if you have a delicious account). Copy the code beneath this, and change CHANGETHIS to your username:

http://del.icio.us/new/CHANGETHIS?
v=2&url=http%3A%2F%2Fwww.brianstorms.com
%2Farchives%2F000575.html&title=brianstorms
%20weblog%3A%20The%20Long%20Tail%20
of%20Tags%2C%20or%2C%20Spamming%20del.icio.us

You will see that at the bottom of the screen are seven keywords (at the time I posted):

spam del.icio.us tagspam via:waxy airset dumb tagology

I would think that most people would choose at least the five first of these if they were interested in the subject.

Posted by: Oyvind at May 27, 2005 06:28 AM

Sorry Oyvind, but I had to slightly edit your posting because that huge URL was clobbering the CSS formatting of this page. You'll need to select the multiple lines of that URL and connect 'em back together into one big huge text string of you want to try out what he suggests...

Posted by: Brian Dear at May 27, 2005 11:24 AM

Hi Brian,

I'm one of the engineers on the Airena team. I think I speak for all of us when I say that we're really embarassed. I haven't vetted this message with Brian Dougherty (sorry Brian), but I wanted to offer my own perspective on this.

Patrick sent us all a note suggesting that we check out del.cio.us. He said it was really interesting, and that in addition to just wanting us to see the site, he hoped that having AirSet appear there would help us to introduce people to our site. Our goal wasn't to spam del.icio.us. Honestly, I figured that we are such a small group that there was just no way we would generate anything relevant, but we did want to at least appear there, and I don't think that's really a bad thing, at least if it's done honestly. (Flame away...)

Let me quote Patrick's email now -- "If you already have a del.icio.us account, please add a link to airset.com url and add a bunch of relevant tags."

Patrick's mistake (sorry Patrick) was failing to consider the laziness of the engineering staff. He explicitly said "a bunch of relevant tags" in his email, but then below he added "here are the tags that I used." Now we're a pretty smart and hard-working group, but I suspect that we all copied and pasted, because it was there, and because we weren't really thinking through the implications of it. I suspect that none of us will make that mistake again.

That's the view from here. Again, I'm sorry.

-john

Posted by: John at May 27, 2005 11:37 AM

This is not the first attempt to spam del.icio.us, and it will not be the last. I worry about companies wanting to spam who find the obvious flaws in Airset's technique, and improve the method.

I also hold a hope that sites like this one will continue to share the results of the time investment required to expose and minimize this link/tag spam.

Mr. Hurley's comment, followed by the one from the employee Mr. Wedgwood containing the actual request, evokes thoughts I had after reading the recent del.icio.us post, Why smart people defend bad ideas. http://www.scottberkun.com/essays/essay40.htm

Posted by: Brett at May 27, 2005 01:58 PM

I'm going to give the airset guys the benefit of the doubt. A lot of del.icio.us users, including myself, use the updated del.icio.us interface when bookmarking. If you go into your *bookmarklet* and add the "/new/" directory before your username

http://del.icio.us/new/[username]

next time you try to add a bookmark, you'll notice a set of "recommended tags" at the top and "popular tags" at the bottom. I could very much imagine a scenario where someone who may have seen the airset bookmarks just copied the tags themselves.

I loathe spam, but asa delicious user, I can kind of see other users just copying other people's tags.

Posted by: Mark Zabala at June 20, 2005 09:31 AM

brianstorms is Brian Dear's weblog. Non-spam email:

Be sure to take a look at these other fine websites:

Copyright 2002-2003 Birdrock Ventures. brianstorms is a trademark of Birdrock Ventures.