-
Website
http://avc.com/ -
Original page
http://www.avc.com/a_vc/2009/07/the-fine-line-between-informing-and-spamming-your-followers.html -
Subscribe
All Comments -
Community
-
-
Top Commenters
-
Popular Threads
All Comments
The huge appeal of aggregation tools like news readers, and even Twitter, is that I can select the channels that interest me, and ignore the rest. But the rest is still available to me, and I can update my feed any time. I really like the idea of having access to your music list (and book list, and blog list, and restaurant list, etc.). I LOVE the idea of combining your tastes with some algorithmic knowledge of my tastes, and exposing myself to a world that is far more open than my narrow selections, but far less random than just picking new things.
The first challenge is to develop tools that can segment a person's output by type, so I can follow Fred and get more or less of his music, industry analysis, business deals, or restaurant reviews. The second is to optimize my consuming applications for the content type. I am not sure I want to see music recommendations in my Twitter feed, but I am sure I would want them to influence my iTunes Genius list.
It makes a lot of sense for recommendation systems to notice that I like Fred's writing style, industry, and taste in books, and to assume that his taste in music is probably relevant to me. But it doesn't make sense to just dump that raw feed in my lap, along with similar feeds from dozens other people, hoping I can find value in the resulting clutter of data.
A landscaper once told me that the difference between a flower and a weed is placement. I think the same is true for spam and news / information. There are very few streams of information that I am not interested in, but if they interrupt me when my focus is directed elsewhere, they become spam.
That being said- I find the people I follow difficult to deal with because they are too tool oriented and not people oriented enough. Tweeting would be more interesting if they had something real to say-
A professor told me and my class that what we put out there matters and hence be selective with what you create, for most of what we as a society puts out is junk. It's annoying to be surrounded by it. But I have to say, that because so much of it is Junk, sometimes evening finding the relevant is extraordinarily difficult.
I wish I could find a list between my two accounts of relevant people to follow if I want to learn about the world in a variety of ways and forms.
It's obvious that companies (especially new companies/startups) use "blind" Auto-tweeting as a tool to build their following on the back of the unsuspecting account holder's reputation. They'll tell you 100% of the time that it's just a viral marketing technique. Wrong - it's manipulative and evil. The most egregious part of it though is their use of the first person in the auto-tweet message, obviously written with the intent of deceiving your followers to believe that you the account holder are making a recommendation of their product/service.
And, they all make the same mea culpa when they get caught. It goes something like this; "We're very sorry, it was a simple coding error" or some bullshit like that. In this case however it was an obvious attempt to manipulate the system to coincide with and leverage their announcement of Sparq at the TechCrunch CrunchUp yesterday in Redwood City. Pathetic.
I hope that most of our users see the auto-tweet as "informing" to their followers, because backing up your tweets can be a good idea if you value them and ever want to reference your old tweets later (Twitter only returns your most recent tweets and Twitter Search only goes back so far).
I think your service (BackupMyTweets.com) is quite interesting and have recommended it to a few people. I also think you need a lesson in emoticon interpretation, because you seem to have a problem with the one that means sarcasm. ;-) ... See, I did it again.
The notion of implementing autotweets because you justify them as "informing" is silly. Imagine if every time you signed up to a mailing list that the listserv automatically sent an email to your entire contact list "informing" them of the list/service/whatever that you had just joined. I think everyone would call that spam, and it is what we'll call autotweeting in a little while.
You shouldn't insist on it. It should be optional. If your service is as good as you think it is, let word-of-mouth grow your business. If you can't survive without the autotweet feature, maybe you should ask Fred to help you with your business plan. And no, that's not me being sarcastic.
Best of luck.
@iboy
Seth's company did something that's unforgivable - sent an auto-message without informing the user and without a choice.
whenever you grant access to any third part to connect with your Twitter, Facebook, etc accounts, it should inform you clearly of what and when they are publishing any content to your accounts, and how you can control that after the "connection" has been done. Almost all implementations fails to do this correctly.
Facebook does this very good, but it also should be third party apps who does that on their own side.
@iboy
Besides, do we really need to develop best practices? When and where did people get the idea in the first place that it's ok to essentially highjack someone else's identity for the purpose of promoting their stuff? Because that's what Blind Auto-Tweeting is; Identity-jacking.
If they're really serious about addressing the issue they'll reverse the process and ask users to explicitly opt-in, which will open the user's Twitter message box and include a pre-written message which the user can amend if they'd like. Now that's a best practice.
For any third-party app that wishes to tweet my action (in their app) to my followers, the notification drops into a 'funnel'. At the top of the funnel is a simple 'This is what we're doing. Are you sure?' dialog, with the standard 'click OK' to continue. The next step actually generates the reply, setting it up for you to click 'Send' or 'Publish'.
I've seen some third-party apps that already use this approach - the funnel is entered in their app & when I choose to 'Tweet this to my friends', it simply opens a new window with the message formatted & ready to go - but I have to click 'Submit' before it leaves the funnel & enters my followers' streams.
Why not normalize this out to a consumable API call? That could conceivably allow me to tune the 'volume' of the notification (100% of followers, only followers I've ever DMed, etc.). That would respond to some of your concerns about how to broadcast / share these messages without being spammy.
These autotweeters are kind of pain. They are killing themselves because when you sign up for tweetapps you expect them to do it and since you can't control what/when to whom they send they are actually hurting themselves. They should flag what they are going to do and let you control it with frequency, target and message filters that pop-up for yes/no approval.
I know its been said before but twitter reminds me so much of early AOL chat rooms. Got to watch that experience doesn't degrade like what happen on AOL. Block goes a long way but need to watch what can be sent to the flow so doesn't reduce to stream of garbage
So, I Alerted Seth. He called me back from the road immediately. He explained that I had indeed overlooked the "checkbox" disclosure but was 100% responsive and took a responsible position for the fact that I didn't clearly see it.
I appologized to all my followers and to Seth and his IDG SF VC for my own fault. Seth has since emailed me saying he is looking into making that disclosure much more clear. I am impressed with Seth Goldstein and his new app, Sparq. With his technology and his attitude... I trust he will go far.
- Bob Wan-Qi Kim.
This just seems to be a decent way to conduct yourself and it shows respect for your users.
Applications which have the power to tweet on your behalf invariably will.
We have all been tricked/trapped/tantilised into signing up for twitter applications which have done the dirty on us.
The major successes in the Facebook economy grew, initially at least, as a direct result of "leveraging your social graph" - also known as "Spamming your friends".
There is no reason to think that app developers in the twitterverse wont attempt the same play-book. Grab land as quickly as possible now by acquiring followers by whatever means necessary and just hang on to them long enough until the monetisation fog begins to clear
Facebook were forced to provide tools to segment your social graph as a result of privacy concerns. Ev and Biz will end up being forced to provide the same tools, not due to privacy concerns but due to spamming ones. We will probably end up being able to segment followers into different groups (business/personal/music lovers/silly video link addicts/bots etc) and each tweet can be configured to which group should receive it.
Not the optimum solution by any means - but one which will become increasingly necessary.
You raise some important style questions. How should we treat our followers and the information within twitter and other open social media. I believe that the tools that add to the noise will ultimately lose out by folks unfollowing spammy streams. The information retrieval tools that learn from our streams and our friends streams without cluttering the data will benefit from rapid user acceptance. Imagine hype stream getting the meta data of your tweets, identifying when you naturally chat about music without a hashtag.
I like the concept of filtering the stream, and customizing the type of messages I receive from folks I follow. This allows me to effectively follow more people, but only select tweets/friendfeed statuses/disqus comments that pertain to topics I'm interested in.
I can use friendfeed for some of that now, but what I'm really looking forward to is real time semantic extraction, and meta data filters of streams (another application of the intelligent search/ad tool I'm going to keep working on thanks to Zemanta's hard work). That type of product will allow us all to become less inhibited by the fear of spamming our followers and lead to a richer and more personalized user experience.
The UI will have to be cleverly designed to allow simple adding of known user tags (derived from previous tweets). This could apply outside of text to images, music, movies or alternative media as long as proper extraction algorithms are used.
It's why most Open sources programs have bad interfaces, but are feature-rich and powerful for the things they need to do. Meanwhile, most closed sourced programs tend to be extremely better designed as long as they act on a well thought out design philosophy.
Thinking of a way to easily semantically tag something that is only 140-160 characters long (They reserve the names, correct?) where each individual character on the human level could have multiple meanings depending on location for a computer to understand is a huge project.
How would the letter c be coded is the following tweets semantically?
"I see you" versus "I c you" versus"Lovely code there" To native english readers and Tweeters who have been there for a while this will be obvious- to newer readers/tweeters, I don't know-and a computer will be totally blind to the difference the way we have them rigged up currently.
The top system has to be easily understood, since the average user does not think, nor knows about the function calls of a program. Their model inside their head is going to be the UI, not the complexity of the fact that computers are blind to semantics and humans are trying to work around that.
If you had to design this- who would take presedence, since the order of complexity is huge for just English?
My plan is to use the best technology that's available in a way that will improve it's natural language and short hand information extraction over time. Zemanta's done a good job of solving the 140-160 characters -> tags and I think they'll continue to do so. The spaces, objects and descriptors help to find broader categories as well as specific keywords.
Giving users a way to help the system learn (memory) will be a strong addition.
Life ebbs and flows and so should your twitstream. Not every day is a perfect six.
Let people unfollow if it's too much.
Would be great if Twitter had a "temporary unfollow."
you could then deal with some spam, or issues you dont care about from a "wholesale" perspective
bottom line is i try to mix it up and i am conscious of that and the volume.
And Fred it speaks volumes about you and your integrity that you would even post this as a "apology" but those who follow you and read this blog would all agree there was no need. You've worked hard to build up this level of well earned respect and one erroneous, unknown tweet won't do jack shit to diminish that.
So how did people get here- who told them?
It reminds me of how Facebook was also going through this phase quite a while ago. You had an influx of several interesting applications, but then they started forcing users to invite all of their friends and it just created disaster. It made for a very spamy experience and led Facebook to do away with that kind of invite-mechanism.
I think the social networking apps that are ultimately successful though are the ones that can to add to what specific users already do every day, be as contextual as possible, and limit the amount of external actions required of them (ie: spamy advertising). Hype Machine seems to be a great example of that potential, as well as others like Stocktwits and Twitpic.
This is the approach I'm banking on for the fantasy sports game I just released built on top of twitter ( http://gawk.it ) ... basically it's a game that scores your sports predictions you log via twitter (something a ton of people already do for no real reason or goal anyway) ... and while there's room for people to start spamming their followers I think the long term players will find the right balance between logging predictions (ie. playing the game) and spamming ...
Anyway I guess my point is if you are going to build something like this, you need to not only be as additive as possible, but fall in-line with existing actions as possible ... sounds like the hypem fits both of those criteria very nicely ...
We just have to be careful of businesses that are trying to push the viral aspect of their products. Let users decide what to share through their status streams.
The constraint at 0.25 (the floor for the friends/followers ratio) is there to still allow certain people at least some voice, while the ceiling at 3 is to limit the amplification of those with a good ratio.
However, this is misleading for people with, say, 200 followers. For example, I have ~300 followers and follow ~200. A bad ratio based on standard algos.
However, I don't follow most of the people that follow me. Giving me a decent 'score' based on attracting people due to the merits of what I say. Seems like an algo that calculates the ratio while eliminating cross-following would help out the 'little guy'. And there are lots of little guys, that when aggregated, might have something interesting to say.
In this case, while it irks me no end when I get spammed or am tricked into sending out spam, I think the underlying issue is on a little different vector.
It seems that while restraint and judgment are required in any editorial activity, in this case we need a bit more systemic approach. Anytime you've got multiple topic areas flowing through the same communication channel you'll always run into this problem - one of segmentation. Allowing the audience to choose which of your channels or virtual streams to tune into, Professional vs Personal vs Stuff I'm trying, would help avoid the spamming issue. Obviously anyone tuned into Stuff I'm trying or the Fred's Music channel would get what they were expecting.
The problem to me seems that while hashtags are helpful, they are inadequate for this purpose without a universally accepted system for segmenting my tweets by channel/topic area and a simpler way for twitter clients to choose among the multiple virtual streams from a single account. In the meanwhile, I'm beginning to experiment with multiple accounts but that is clearly not ideal.
Any other thoughts on creating virtual streams?
main feed - http://feeds.feedburner.com/AVc
vc/tech feed - http://feeds.feedburner.com/AVcVentureCapitalAn...
music feed - http://feeds.feedburner.com/AVcMyMusic
it uses the categories of the posts i set in typepad
joshua schacter showed me how to create custom RSS feeds by category in typepad (we hacked it)
then i ran them through feedburner.
now, i am not excited by self tagging my tweets. but i love the idea of zemanta and calais doing it and then allowing users to follow me by tag or a set of tags.
someone will build this. maybe they already have.
TCP/IP, HTTP, twitter, zemanta, <tag processing>
With regard to the Hype Machine's Twitter list, I understand what you are saying about it having "game dynamics," but at the same time, I certainly hope that people DO NOT start thinking of it as "a game." Anthony created the Hype Machine Twitter list, just as he created Hype Machine proper, as a tool to help people discover the music that others felt passionate about. It's a social tool created to provide a real value, and as such needs to be founded on true, accurate data. Hype Machine (regardless of what the name may imply) was NOT created as a tool for bands or PR people to use to game promotion, nor was it created a means for blogs (or now, Twitterers) to brand-build for themselves. Exploiting Hype Machine like this lessens its value to *everybody*, and is the root of Anthony's recent problems with Hypem "spam" detailed in his "On Chart Integrity" blog posts ( blog.hypem.com ). In the same way that Black Hat SEO tricks devalue the search experience for everyone, so blogging/tweeting a song from a gaming/promotion motivation devalues Hype Machine entirely. I don't think there's any sort of fine line between spamming and informing-- it always seems perfectly obvious to me which is which.
I think that those of us who grow to wield more influence than the average online Joe need to be much more aware of what we're posting and why. We should be adding value to the online experience-- and not value for the one (the one asking you to post/tweet), but value for everybody.
hype machine is creating this game. in my opinion the burden is on them to ensure the game is managed the way they want it to be. if they don't like the incentives they are creating, than they should simply not create them.
The use case I have for it is that there some people on planet.mozilla.org whose personal, non-Mozilla postings I want to see, without seeing all the duplicate Mozilla articles that I've already seen on planet.
In theory it sounds great; the current sticking point that I'm hitting is that you generally have to stuff the feeds through Yahoo pipes or some other system that has boolean logic in order to get the desired filtering, which ends up making the UX more hassle than it's worth.
My current suspicion is that a feed-reader with built-in (tag-based) faceting UI for adjusting individual feed content could do pretty nicely.
I also created separate RSS feeds from topic areas on my blog to give people more choice on what streams of interests they want to follow; in the end they are all hacks to address the problem that Peter Laudenslager describes in his comment below, of a system where we follow people rather than just facets of people.
I`m not sure if people actually care that I segmented out facets of me, but it makes me feel better :)
(Heh, and now I'm checking the Disqus option to tweet this comment. Permission is everything, especially when software can "speak" on my behalf....)
That's a phrase to sit and think about
Well put
I get that there's a trade here. But how that trade is executed is a big deal
2. Re: hypem, I think it's pretty simple. Be selective and only post when you really love something, not on a forced schedule or frequency, and don't worry about being spammy.
Looking forward, I like the idea of tagging/categorizing/filtering tweets, but it requires work on the sender's end, the receiver's end, or most likely both.
Here's one half-baked thought that just crossed my mind: "Twitter App Streams". Apps like spymaster would send all activity tweets via their own account, on behalf of the player (which would require oauth / be an attribute on the api call). Then, only people following spymaster AND the player would see those tweets. You'd have invitations and announcements outside of this, but it would keep the heavier 'in-game" traffic filtered to only those interested in it.
Another simple* alternative would be to keep things how they are, but require all applications to register with twitter and authenticate all their tweets, then provide some basic filtering to let the follower squelch certain apps. If everything is opted-in by default, it would work just like today, except you'd have to ability to filter out noise without having to unfollow someone who you otherwise enjoy.
* - Trivializing a hard problem. The concept is simple, the implementation I'd imagine is far from it.
thanks for this post.
:}(o|O){:
At uberVU we are indexing everything and what we discovered is that top results are heavily influenced by spam. Thousands of tweets containing poplar words or popular hashtags is how they do it. I don't know if Twitter should fight spam the way blogging platforms did or it should be the developer's job. But if Twitter wants to be more than just a personal tool for following friends they will need to address this asap.
At this point anyone who thinks forcing auto-messaging might be OK is clueless or is OK building their business using spammy techniques. Plaxo made the downside of it clear years ago. It's a poor and poorly thought of practice and you do it at the risk of your company's reputation.
A key example of this is live tweeting events. When a major event like SXSW or Web 2.0 comes along, the tweet stream gets flooded with live tweets on topics I'm not interested in. I don't want to unfollow these people, but I need filtering tools to get rid of those tweets.
I've thought about creating a separate Twitter account for people who want to follow my travels, but that's more work for me and my followers.
The sooner Twitter can allow such filtering the more engagement in generate (people who are interested in those niches) and the more data that can be collected for analytics and search.
More thoughts here:
http://blog.agrawals.org/2009/06/03/to-tweet-or...
If I could make a suggestion - make it mandatory to disclose any automatic actions an app will take during the signup workflow at the oAuth authorization screen. For example, an Iran election application overwrote my background and set my location and timecode for me. The background change was not disclosed. There was no option to opt out or edit an automated tweet.
If an application posts a tweet that I can't edit, I will either not use it or immediately delete it.
I want to see "An application "XXXX" by XYZ Inc. requests permission to access your Twitter account to read and post updates. ADDITIONALLY, it will post "Promo tweet text goes here" that is EDITABLE and OPTIONAL, and CHANGE [Profile, timezone, bio...]" right on the oAuth initial authorization screen.
Obviously, non-editable and mandatory tweets are red flags. That would decrease the amount of spammy apps.
"I'll be the first to admit that being an investor in and a board member of Twitter has helped. But I've not wanted to be on the Suggested User List and I am not. The people who follow me on Twitter have chosen to follow me for a reason and I try hard to post things that they'll find interesting."
A good example of how this could be solved is how we implemented the Spymaster filter with Topify.com that blocks any incoming direct message mentioning the game by just sending an email request. This has been a very popular and appreciated method.
http://blog.topify.com/post/116286293/no-more-s...
Twitter should simply do this.
I don't necessarily agree that there is a "twitter limit" of 4 to 6 messages a day: it really depends of the use you make of it. If you use it to inform, that's probably right. If, however, you use Twitter as a communication media alongside, and maybe instead of, email or phone, then you might be Twittering much more, and then people will decide if you are spamming or not. For example, I use Twitter to talk game business, to stay in touch with friends, to crack jokes, to describe cool stuff I'm doing, to share interesting links and to inform (v rarely) about products or initiatives that PopCap is launching. Overall, I probably send 15/20 tweets a day, and some people might be interested only in some of them, but I'm not feel like I'm spamming: in fact, it happened to me that people that followed me for game industry weets commented to me about how they enjoyed my food-related tweets, or how a joke I made cracked them up. Same has happened to me countless times, following someone for one reason and finding other cool stuff through that person. I think that's cool, and while it's more serendipitous, it's one of the most intriguing features of Twitter, which otherwise would be much more stuffy, boring and "on message". That said, I also came across people that tweet moderately and that I consider spammers, because they only self-promote (several celebs are in this category), no matter the volume of their tweeting.
It's not about the number of tweets, it's about how funny/useful/interesting they are.
Cheers,
Giordano
Shaq, for example, could tweet forty times a day and I bet his followers wouldn't mind
The tension you feel is what makes the chart so interesting and proofs it against spamming, glad you like it!
Even in yesterday's example where Techcrunch attempted to rickroll the chart by tweeting a Rick Astley song (it went to #1, because of their 900,000 followers), it only stayed #1 for a little while until Mashable tweeted a MJ remix, putting it back to #1 (with their 1.1M followers).
The system auto-corrects socially.
People on Twitter sometimes forget they are people. They become advice and brand pumping machines. To the extent they allow a bit of their personality into their posts - that's when I choose to follow. The trick is to not let the game, or the brand, or the guru-ness overtake the real person. I like to learn. I don't enjoy being beaten down.
>> I think that's true. But my tweets aren't really for the followers, they are for me, it is my online diary, where I post links to news stories, to songs I love (via @hypem or @lastfm, etc) ; where I vent, where I rant, where I let my co-workers know where I am at. I hear you there, but it really depends on what you use Twitter for .. I'm not trying to win a 'followers' race.
>> Also re: spamming. Side topic. I'm spending $1.50 per added fan on Facebook. So let's say my CPC is $0.50 on Facebook to advertise my band page. 1 out of 3 clicks leads to an added fan on my Facebook fan page. I like to thank new fans who join by messaging them (directly) a folder of free tracks. After sending about 5 notes to 5 new fans, Facebook informed me I was 'engaging in spam like behavior' and that my account could be suspended. I don't get it. I paid for access to the fans and now I can't directly message them? I love that I can hyper target potential fans and convert a few into real fans using Facebook advertising. But what's the return for me on this advertising investment, if Facebook is going to limit how I can contact these fans.
Regards from SF, Chuck
http://chuck.extendr.com
As entrepreneurs, there is tremendous pressure on us to grow. But there may even be a business reason (besides the ethical reasons) for not tricking people into sending spam. If we don't regulate / police ourselves, someone else will come in and do it for us. See the Tagged.com situation.
As such, it seems that Twitter (or someone in the ecosystem) could cull tweet data to come up with a user's reputation score based on things like tweet usefulness, accuracy and domain relevance (context) to better guide users. Plus, this same data could be used to expose "handles" so client apps can filter tweets and message payloads accordingly.
The analog that comes to mind is the Whuffie, the reputation-based currency concept in Cory Doctorow's Down and Out in the Magic Kingdom: http://en.wikipedia.org/wiki/Whuffie
Taken together, when you combine Verified User with a Reputation Score, you have some potent currency, and a huge impetus for users to self-moderate their approaches to communicating with followers.