Recommending People Without Being Asked

by Adam Stallard

Loading mentions Retweet

So I've decided to start a new practice for myself.  Every day I'm going to go to my account on linked in and write a recommendation for someone in my connections; and I'm going to do this without making them ask me first.

I think the toughest part about getting recommendations is worrying that you might be putting someone out.  Well, now no one has to worry about asking me.

If any of my six readers decide to do this too, please let me know, and let me know how it goes.

Heb 13:16

by Adam Stallard

Loading mentions Retweet

Believe that communication is worth it.

But to do good and to communicate forget not: for with such sacrifices God is well pleased.

Don't consider this a gospel sermon, consider it just another post in an altogether dogmatic blog.

I think when we think of sacrifice, we focus on what we lose.  Sacrifices of a biblical variety are different: they involve giving up something now to get back something even better later.

Using this biblical definition, how is communication a sacrifice?  Communication takes up your precious time.  It can be uncomfortable--and you when you open your mouth, you may run the risk of looking foolish or offending someone.  But you get something in return that more than offsets what you gave up.

You get feedback.  Or for those kind-hearted souls who understand the ethic of reciprocity and want to be proactive in helping out their fellow human-beings, you give feedback.

People want to change and be better; and they can't do that without feedback.

Do you remember the kid in elementary school who liked to show off his pictures and wanted you to tell him what you thought--but only if it was nice?  Unfortunately, that kids' pictures never improved past a 5th-grade level.  (Or more likely, someone eventually burst his bubble and he actually started getting good.)  But what's even more unfortunate is that no one wanted to talk to that kid because they felt uncomfortable about it.

Communication Avoidance

It's hardest for me to communicate when the person I want to talk to is expecting something positive, and I know the truth is negative.  I go through an internal conflict between wanting to be honest and wanting to be positive.  To avoid feeling the displeasure of that internal conflict, I avoid communicating.

But communication is one of those sacrifices that's always worth doing.  This has hit me the hardest at those times when I've been on the receiving end of non-communication.  I remember thinking, "I just want them to tell me something, anything--the silence is killing me."

We're afraid of disappointing someone, but that's foolish.  Withholding information is almost always a disservice because it interferes with people's ability to act. If everyone goes around hiding the fact that someone did or is doing something poorly, they are seriously hurting that person's chances of doing better.  I need to show some faith in people's ability to learn and change--every single person deserves that.

Being upfront offers everyone involved the chance to make the best of the situation.  On the other hand, things not working out the way we expected plus silence always equals things not getting better.

A Resolution

So here's my resolution:  no more thank-you notes or phone calls that never get made.  (Even if I'm struggling with a way to tell someone I'll never wear their sweater, they deserve to know that I appreciate the thought and it didn't get lost in the mail.)  No more avoiding someone because they want a recommendation letter and I don't think they deserve it.  No more avoiding a client because I don't have the heart to tell them that I want to spend less time on their project and more time on something I find more valuable to humanity.  No more giving someone the silent treatment because they've hurt my feelings.  No more stringing people along because I'm afraid of telling them I've already made up my mind.  If we hope to get better, we have to communicate these things.

It's all about feedback.

If I want it, I have to be willing to communicate.  That's the essence of the sacrifice.

Self-Reminders to Work Smarter

by Adam Stallard

Loading mentions Retweet

Be more efficient

1. Don't do today what you can put off until tomorrow.  Tomorrow there will be fewer unknowns, better tools, and more money.  Can this be decided later?
2. Your time is valuable.  Is there a way to do what you're doing without spending time?  Can someone else with cheaper, more available time do it?
3. You learn more by doing than by researching.  If you come across a solution that could fit your requirements, try it and see.
4. Is the thing you're doing supposed to save time?  If so, is the time you save greater than the time you're spending implementing it?

Hop on Pop and Language Discovery

by Adam Stallard

Loading mentions Retweet

Learn how we all share Dr. Seuss's instinct to invent and discover language.

Welcome to another installment of Trending Tuesday, where I pick a trending Twitter theme and relate it to goal-based communication.

I'm going to take this opportunity to pay homage to one of my favorite childhood authors and a man with whom I share a birthday:  Dr. Seuss.

The Simplest Seuss for Youngest Use

I've been reading books with my two-year-old niece lately.  Or, she's been letting me read--she doesn't seem to grasp the concepts of words or letters yet. She can recite to you her favorite books using the drawings as cues,  but since English isn't composed of pictograms, she's unlikely to learn to read or write (or even notice words) this way.

There is one book that stands out in her collection: Hop on Pop.  This is an incredible book and I'll tell you why:  the author draws attention to the fact that we're using words. He does this with the most blatant rhymes imaginable. These aren't pretty couplets or Poe-esque meters; they are in-your-face phonetic assaults.

Here's a typical page:

HOP POP
We like to hop.
We like to hop on top of Pop.

When you read that, you feel like you want to exaggerate the sounds. Because the words are written very large, my niece can easily see which one I'm pointing at as I pronounce it for her.  I'm hoping that soon she'll say the words herself as we point to them.

What was so great about Dr. Seuss?  He loved language--he loved the way words sounded and got us excited to say them.  In a time of prettily-painted story books, he wasn't afraid to write in large fonts and make words central.

This Stuff is True for Grown-ups, Too.

The process of learning a new language applies to everyone.

Every time we meet someone new, we reinvent how to talk. We have a different language for each person or group we know--it's a linguistic contract that's developed in stages.  Communicating this way involves testing out different pieces of language to see what's comfortable and agreeable and convenient for creating shared shortcuts.  Like a child learning to read, we set up and discover what words, phrases, and physical expressions will mean.

In linguistics we learned about idiolects--how every person has their own unique way of speaking.  But that doesn't capture the fact that I have a different language for each person I talk to.  To me, it makes more sense to talk about duolects: the class of unique languages shared by me and exactly one other person or group.

In a future article, I'll talk about small talk, why it's so important in the development of a duolect, and why we need software that can simulate small talk and get it right.

Good Software Principles and Practices Can Wait

by Adam Stallard

Loading mentions Retweet

Learn what works best as you go.

I'm excited about my current team; individually, we have a lot of experience with a variety of software practices, so I feel like we have a good sense of  what does and doesn't work for a given context.  I thought it'd be fun to have a brainstorming session to talk about what design practices we've had the most success with or were itching to try out for our current project,  so I called a meeting with my team (all two of us) and we started listing things off.  The ideas started getting long, detailed, and speculative until my teammate said, "this doesn't seem agile anymore."  We decided to stop right there, not make any decisions, and just go to work and find out what we actually do in practice.

Having learned from this, I'm going to do things differently now at the beginning of a project.

I'll make two sheets of paper (or wiki pages).  On the first sheet I'll write "principles".  Principles are ideas that aren't language or implementation-specific like "Make production code readable and maintainable".  On the second sheet, I'll write "practices".  These are more specific, like "use object oriented CSS".  They also change more often and no one is ever forced to use them.  We want to hold the team to principles, but practices are more often just a catalog of what we're doing or trying to do.

Now that I've made these two sheets and I know what they'll be used for--I'll leave them blank.

They stay blank until after the first sprint, at which point, there is a short post-mortem where we write the things we did and liked under "practices."  Then we write just barely enough words under "principles" to stop things from going wrong again (having previously done on-the-spot root-cause analysis).

I think this more in line with the agile principle of "You Ain't Gonna Need It" (which I mentioned in a post on the merits of procrastination) and I feel good about it.

What Twitter Ads Should Look Like

by Adam Stallard

Loading mentions Retweet

Find out what makes a good advertising platform.

Twitter is coming out with a brand-new ad platform next month, which means they have a chance to start from scratch and do a lot of things right.  There's been a lot of speculation about what Twitter ads will look like.

What is the single most important criterion for determining a good ad?

I think it's that users only see it when they want to.

I don't mean being able to close a pop-up; I mean saying, "I feel like seeing some ads right now" and then having them show up. 

How do I know it's important for users to only see ads when they say so?  Because as soon as people discover that they can get rid of ads they didn't ask for, they do it.  They record their t.v. programs and start them 20 minutes later just so they can fast-forward through the ads.  As soon as they realize they can use adblock and flashblock to block internet ads, they do it.

At the same time, they voluntarily watch videos of ads they've heard are funny or creative, and share them with friends.

Why would anyone choose to see an ad?  They would if it had similar information to what they're seeing right now, but better.

Ads nowadays have become targeted, meaning they get half of this equation right.  They have gotten pretty good at nailing down the "similar information" part, but they totally fail when it comes to "but better."  You can pretty much count on information that has "sponsored link" next to it to be anywhere from slightly to incredibly lower quality than information that doesn't.

As I said in an earlier post, Facebook ads are pretty cool from an advertiser's perspective because they let you select your target audience based on tidbits from their Facebook profiles.  But how do advertisers know if they have the right audience?  If they're smart, they'll keep trying different combinations of profile data and measure their success at different points along the customer conversion process.

Once an advertiser has pinpointed their ideal audience and knows exactly how the profile of their best-performing customer reads, they can just show the ad to everyone with that profile, and they'll all be happy about it, right?  This is the secret to "but better," right?

This may come as a surprise to the advertiser who views human beings as a list of profile attributes, but what people are interested in changes from moment to moment.  The compliment to knowing the answer to "who is my audience?" is knowing "what are they trying to do right now?"

Armed with "who is my audience?" and "what are they trying to do right now," advertisers have a chance of tweaking their ad all the way to the "but better" range.  In a scheme that only shows ads that people want to see, you have to be in the "but better" range, or you just don't show up.  Fortunately, there are so many combinations of audience and objective that anyone can find one that works for their product or information--if they have access to the right feedback.

The key for Twitter lies in the fact that "what's happening?" is often only a short step away from "what are you trying to do right now?"  If tweets can be filtered for information about people's current objectives, then advertisers can be supplied the data they need to be able to know when it's the right time to show an ad.  If this happens, then Twitter ads could actually be a good thing.

Prisoner's Dilemma and Goal Alignment

by Adam Stallard

Loading mentions Retweet

To illustrate the power that comes from engaging in early-stage communication and goal alignment, let's turn to the prisoner's dilemma.  The problem is described like this:

Two conspirators in a crime have been caught.  They are held in isolated rooms for interrogation.  It is explained to each one separately that if both confess, they will both serve 6 months in prison.  If one of them denies involvement and the other confesses, the confessor will get 10 years, and the denier will go free.  If both deny, they will both get 6 years.

As a conspirator in this situation, what should you do?  If you can't predict what your co-conspirator will do, your best bet is to defect.

The problem here is that the conspirators can't communicate.  If they could, they would quickly zero-in on mutual confession as the best solution.

Wired magazine details what happened in 2004 when the paradox was computerized and transformed into a team competition:

The Iterated Prisoner's Dilemma is a version of the game in which the choice is repeated over and over again and in which the players can remember their previous moves, allowing them to evolve a cooperative strategy. The 2004 competition had 223 entries, with each player playing all the other players in a round robin setup.

Teams could submit multiple strategies, or players, and the Southampton team submitted 60 programs. These, Jennings explained, were all slight variations on a theme and were designed to execute a known series of five to 10 moves by which they could recognize each other. Once two Southampton players recognized each other, they were designed to immediately assume "master and slave" roles -- one would sacrifice itself so the other could win repeatedly.

If the program recognized that another player was not a Southampton entry, it would immediately defect to act as a spoiler for the non-Southampton player. The result is that Southampton had the top three performers -- but also a load of utter failures at the bottom of the table who sacrificed themselves for the good of the team.

What's interesting here is that the winning team encoded the confess and deny actions into a language they could use to establish a mutually accepted goal (one conspirator would take a rating hit to boost the other conspirator's rating).  In other words, they considered the advantages of communication so great that they invented a way to do it a context where there was none.

Their strategy needed 5 to 10 rounds of information exchange before the goal could be settled.  Just like these computer programs, people need several feedback rounds to align goals.  Too much of what passes for communication on the internet is essentially one round: someone posts information and someone else anonymously reads it.  Consequently, a goal is never agreed upon, and the parties are disappointed.  For a content creator, a process of setting a goal, getting feedback and revising the goal needs to be in place.  For the content consumer, there should be a goal attached to the information they are viewing telling them what they can expect to achieve by using it, and they need a way to report their experience.

Like the Southampton team from the tournament, GBC is introducing goal-aligning communication to contexts where it is currently missing.

Was This Helpful?--Thoughts on Microsoft Support

by Adam Stallard

Loading mentions Retweet

What kind of feedback data do you really want?

Help articles on support.microsoft.com have titles like: "Windows 7 activation error: invalid product key" and "What are the system recovery options in Windows 7?"  The first title spells out very clearly what the user is trying to do: activate windows 7.  The second article covers more usage possibilities.  Failing to distinguish between use-cases in a feedback system can cause failure in aggregate rating calculations (for instance, counting all the helpful and not-helpful votes).

Microsoft support articles all have the same method of collecting machine-readable feedback about the performance of an article.  The user is asked "Was this helpful?" and can choose between "Yes," "No,"  and "Somewhat."

This approach probably works well when an article's title is conspicuous and highly specific--and when users start at the support.microsoft.com home page and descend the well-organized hierarchy of help documentation.  These users are likely to quickly zero-in on the most appropriate document.

But what about a more common occurrence: a user reaches a help article through an internet search engine?  The chance decreases that what a user is trying to do matches what the article's author envisioned.  Users that find an article this way might even skip over the title (which often tells exactly what the article is trying to do) and focus on the terms they searched for instead.

In all cases, when the author gets feedback about the helpfulness of an article, they should know (not predict or guess) what the user was trying to do.  Even if there is a well defined title telling users what they were supposed to be trying to do, and a hierarchal guide to direct them to the best article, they still might try to use the article in another way.  Why fight this? Why not let the user determine what they are trying to do?  Let them rate the article, but attach meta-data that indicates what they were trying to do--even if it's different than what the author expected.  Encode this meta-data in a way that ratings can be automatically sorted according to use-case.

What this accomplishes:

1. People trying to use the article for something other than what was originally intended can be quickly directed to something that does what they want.
2. An author might find out that their article is useful in ways they didn't intend.  New ways to find this article can be added in appropriate places.

Sure you could just read through all the reader comments, but making a rating that can be encoded as meta-data has two advantages:  first, it usually means there is a way to make it quick and painless for the person giving feedback (meaning you will get more feedback), and second a computer can tally the data so you don't have to.

How Five-Stars Fails

by Adam Stallard

Loading mentions Retweet

Learn what should be encoded in a good aggregate rating system.

In a previous post, I said that product endorsements need user-supplied meta-data to make them better.  I theorized that the following two pieces of meta-data are especially important:

If I am reading an endorsement I need to know:

1. The endorser has specific qualities that make the product especially suited to them; I also have those qualities.
2. The person was able to do a certain thing with help from the product.  I would also like to do that thing.

Personal Traits

Should we submit traits such as our age, sex, and location when we're rating the usefulness of a piece of information?  I can imagine situations where it would make a difference.  If we are willing to share those traits (anonymously), and it is determined through analysis that there is a statistical link between one of those traits and the perceived usefulness of a piece of information, that link can be used to predict whether it will be useful to other people with that trait.  Filters based on this data can direct people to information that is more likely to help them.

Should we carry around a profile that automatically gets applied when we rate products or information?  I think so; and I think it works best when we know that's what the profile is for.  Facebook uses a profile that is ostensibly for telling about yourself to your friends.  What isn't obvious is that it's also being used to determine what ads you will see.

Facebook also lets you give ads thumbs up or thumbs down, which is good; but it does not connect the thumbs up or thumbs down rating to your profile traits to help determine what ads other people like you will see.  Instead, advertisers have to guess at the profile of their target audience--never having access to the thumbs up/thumbs down data. A key statistical link is lost.

Separate Use Cases for Better Rating Systems

The second piece of data that should be encoded is the effectiveness of products or information in helping people accomplish particular tasks in real life.

There is a feedback loop involved.  It has four phases: first, someone attaches a theory to the product or information surmising what it will help people do (several of these can be operating simultaneously); second, people try to use the product or information to do something that matches one of the theories; third, they report back their success or failure; fourth, a theory's score is adjusted.

It's easy to find processes similar to this on the internet, but they are missing a key component: the theory doesn't define what a person is trying to do.  (I'll talk about how the quintessential Microsoft feedback system fails in this regard in another post.)

A major failing of aggregate rating systems today is that they don't distinguish between different kinds of attempted tasks. A rating that comes from someone trying to use a piece information to have a good laugh is combined with the rating from someone trying to use the same information to complete their science-fair project, and in the end the aggregate rating doesn't mean anything.  We have to read the comments on the reviews one by one to get meaningful information.  This can be very time-consuming and defeats the purpose of aggregate ratings.

Some information may have a very clearly defined task associated with it; an example of this would be an instruction manual.  But what if the information turns out to be useful in an unforeseen way?  If someone discovers that an instruction manual for painting a wall is also useful for painting furniture, there is no way to distinguish between the different applications in existing rating schemes.  We have to go with user comments again.

If we look at what people put in their review comments, we get an insight into what meta-data is missing from the associated aggregate rating function.  Do people mention personal traits (age, sex)?  Do they mention the fact that they actually used the information or product and what they were trying to do with it? They write these things because they instinctively know they are important to a good endorsement (or anti-endorsement) and one to five stars can't communicate everything they need to say.  I believe that some of this passed-over comment data can be made into part of the aggregate rating system.

Conclusion

Encoding user feedback into meta-data means being able to automate the review process (by making helpful aggregate functions), which avoids the tedious processes we go through now of reading through thousands of individual reviews; but the aggregate function needs to be better than one to five stars.

Permanently Remove Google Buzz

by Adam Stallard

Loading mentions Retweet

In the upper-right corner of your Gmail page is the word "Settings". Click on it, then click on the "Buzz" option (as shown above). Click "Disable Google Buzz" and it will step you through process of deleting and permanently disabling Google Buzz.