Older Posts That Are Still Interesting

Here are some short posts (long in aggregate) that I wrote on a throwaway blog on Posterous around 2009. Posterous got acquired by Twitter, and I didn’t like it all that much anyway, so I exported and sorted by interestingness. Fair warning: they are a bit rough, but might be useful.

How to Figure Out If You Should Get More Education

I was talking to a woman in the information security field last night. She talked about getting more certifications a few times. I say to her, “In software, a lot of times certifications are kind of a joke.” Meaning, a lot of certifications are merely memorizing the documentation or don’t really show that the person knows what they are talking about. She said that many of the certifications were useful–but not without the proper experience to back them up. When she was hiring people, she would first sort through the resumes to find people with the certifications that made sense, and then approach those people first. The fallback was the people with fewer certifications. However, she wanted to see if other people thought the same way.

Next, she said something that I found very interesting. She said that she created two resumes for herself and sent them both to a few companies in her field. In both, she used different names. In the first, she listed her current skills, experiences, and certifications. In the second, she did the same but also listed the certifications she was considering going for. If the second resume got many more responses, it was clear that the certification had market value.

I thought this was an effective approach because she got information before needing to actually take the time to pursue the certification. At first, I thought it was similar to split testing, but what made it interesting was that she was not trying to optimize some asset she currently had. Rather, she was optimizing for future actions. It reminded me a bit of lean startup techniques, in that she was testing a value proposition rather than building the product first and seeing who would buy it later. For something that might take a couple of months of nightly study, it seemed like something worth testing first. I’ve long thought a useful technique has been asking the question, “What are we trying to learn, and how might we learn it faster?” This is undoubtedly orders of magnitude more effective in terms of ascertaining the value of a certification.

One of the signals a certification gives is that the person is willing to stick with something long enough to get it, and also has an interest in doing something in that area in the future. So it can be useful, but not universally. There were some people with a lot of certifications but only a year of experience. These people would likely need to work for a bit to get more broad experience before they would be able to effectively use their certification knowledge.

Clearly the market value of some educational experience is not the only thing you should consider. Is it going to make your career more interesting or add meaning to your life? But these are questions for another time. I hope you enjoyed this technique.

For more on split-testing in the physical world…

Split-testing Online Profiles

Occasionally while going through personal writing, I come upon an idea I wrote down in the heat of the moment and that seems good later. I have a Vim function that prints out the date, so that’s how I start personal writing entries.

20091210 - 0934

I had a pretty powerful chain of connections this morning while washing out my oatmeal. Basically, I thought about split-testing dating profiles to optimally recruit candidates. To get the best results, these profiles could be split among various cities to get normalization of the values. For example, using the same profile, but in different cities to see how people differ geographically. Another would be to have the same profile with a different picture to see what the effects of the picture have on the responses. You would then see the types of profiles or people that respond to your profile, and tweak these to be closer to what you want. Basically this is a mechanism to increase the chances that you are communicating what you want to communicate to other people.

The “whiz-bang, throw my hands in the air” realization is that this could also apply to companies that are recruiting. This was basically the fruition of a month of background processing. You could create multiple profiles on different sites and even on your own main site and see what result using different messages have on the quality of the candidate pool

So this is a way of understanding what portions of your communication are working effectively, and where you can improve your signaling. If I split test interests and “liftin’ weights” gets a better result than “weightlifter”, then I have more insight into the funnel process that I’m working on. Is my call to action effective for the people that I’m trying to attract? Do I really understand their problems and where they are coming from, or am I focused solely on myself? One problem with this process is that the results are pretty subjective. Hits might not be quality hits, etc.

This was probably inspired by OKCupid’s blog and reading stuff about Lean Startups.

Blackhat Twitter Technique Thoughts

As this is my off to the side blog for exploring controversial or experimental ideas, I figured I’d post here. Here’s a potential technique for getting Twitter attention that I’ve been thinking about for the last couple of weeks. I haven’t actually tried it, but it would seem to work well based on a few Twitter mechanics. It relies on the following things:

All About @ Replies

First, @ replies at the beginning of a tweet are filtered out of people’s tweetstreams who are following you but are not following the person with whom you have replied to. For a while in Twitter, you would see all @ replies. However, this was equivalent to listening in to conversations that you had no interest in, so it was annoying. Nowadays, Twitter conversations (at least on the official Twitter clients) that are between two people that you follow show in your tweetstream.

Examples to summarize: if A follows B and C but not D, A will see the following tweets in their tweet stream (timeline):

  • B tweets: this is a normal tweet
  • A tweets: Do I see my own tweets?
  • C tweets: @B hey B, what’s up
  • B tweets: @C hey C, not much, lolz

However, A won’t see any of the following tweets: * D tweets: this is my tweet that won’t be seen by A or anyone not following me * D tweets: @C yo C * C tweets: @D ahoy D

Hashtag Hijacking with filtering

Next, people at conferences often continually scan the conference hashtags out of boredom, to explore the backchannel, and to join up and have meaningful experiences with people that are interested in the same things as they are. Twitter Searches for these hashtags will return any tweets with those hashtags in them, including ones that start with @ replies.

We can use this mechanic to effectively filter the range of people that see the tweet to only people who are searching for the hash tag. Example:

You tweet: @someonewhodoesntexist this is a great post related to the conversation we had http://myblogpost.com #conf2011

No one follows @someonewhodoesntexist so any followers you have won’t see the tweet. However, anyone searching the tweet stream for the hashtag #conf2011 will see the post anyway. Bonus points for misspelling conference luminary names or just using them outright to get RTs and social proof in the hashtag stream.


What you get is a hyper-focused stream of likely early adopters or people interested in your product area to show early products or ideas to. Often people who are not even at the event but are still interested in the proceedings will be tuned into the conference hashtag “channel”. The interesting thing is that it’s not spam in the conventional sense of messages to one account or to a mailing list. This probably wouldn’t work well for established products.

Note that people viewing your profile directly can still see the @ reply-prefaced tweets. In order to not be marked as a spammer, it would help to have tweets at least tangentially related to the conference proceedings. If it’s an iPhone dev conf, your iPhone app beta signup might be a good thing to tweet about. I could see using various fast prototyping events (Startup Weekends, perhaps) to get tweets and views for products that are early stage. Creating an anonymous or one-off account to do this would probably be effective if you had an existing brand.

Of course, figuring out what conferences are out there and which would be most useful is a non-trivial exercise (based on the tools that I know.) Also, tweeting regularly in this fashion would be a manual process unless someone writes tools to do it. Factors of a tool would probably take into effect hashtag stream velocity to automatically tweet (as the stream is ephemeral, so repetition can be effective if not done too often.)

Obviously I’m not advocating this ( blackhat was in the title :) ), but it seems like an interesting hack based on how Twitter is actually used. What are your thoughts, and have you seen this in action anywhere?

Marriage as Technology

Mating equality

Having people pair off in monogamous relationships ensures that everyone can mate. In contrast, polygamy means some men will fail to find wives, and promiscuity results in mating inequality based on attractiveness.

“Marriage is an institution; it places artificial limits on women’s choices. To repeat: Nature dictates that males display and females choose. Monogamy artificially strengthens the male’s position by insisting that 1) each female must choose a different male; and 2) each female must stick to her choice. Monogamy entails that highly attractive men are removed from the mating pool early, usually by the most attractive women. The next women are compelled to choose a less attractive mate if they wish to mate at all. Even the last and least of the females can, however, find a mate: For every girl there is a boy. Abolishing marriage only strengthens the naturally stronger: It strengthens the female at the expense of the male and the attractive at the expense of the unattractive.” “Marriage, like most useful things, was probably invented by men: Partly to keep the social peace, partly so they could be certain their wives’ children were also their own. The consequences of marriage must have appeared soon after its institution: the efforts previously spent fighting over mates were replaced by strenuous exertions to provide for, rear, and defend offspring. No doubt surrounding tribes wondered why one of their neighbors had recently grown so much stronger. When they learned the reason, imitation must have seemed a matter of survival.”

The Wikipedia article on the value of monogamy leads me to wonder if marriage can be thought of as a long-ranging technology with adoption curves? This thought-process was suggested by Venkat’s post containing a concept that the notion of a “firm” is a technology with a hype cycle. This might imply that marriage as a technology has certain ranges of usefulness and adoption rates. Upon re-reading the excerpt, I notice the “like most useful things” of the last paragraph seems kind of strange. Perhaps it’s a Wikipedia openness artifact, but it seems to imply that “most useful things” were “probably invented by men”, which I don’t think makes sense. So it would be interesting to examine the source more closely.

What Breadsticks Taught Me About Negotiation

At a local sandwich shop they have a pizza/breadstick/dipping sauce/drink combo that I fancy. Yesterday I went in there and there weren’t any breadsticks, but I bought it figuring they could make some more. I asked if breadsticks came with the combo, and the guy behind the counter said yes and he looked around for a bit before going to the back.

He came back with a woman I had seen there before often and said that they were out of breadsticks. He offered me a refund. I kind of thought this would be a pain since I paid in cash. He then quickly offered me the last remaining personal-sized pizza and I thought that it sounded like a lot of food, but said yes. The woman smiled at me, and then the guy said thank you for working with us. He then said that they always give the last pizza away for free anyway. Once I said yes they quickly directed me to where I could find the bonus pizza. I didn’t have much time to change my mind.

This last sentence he said was an interesting development because I would have felt like they gave me a better deal if I didn’t know that they always gave the last pizza away. Maybe he said that to make me feel better about our transaction. All in all, it would have been more effective to not tell me what his costs were like.

The setup suggested that an effective strategy for small coercions is 1) offer something mildly unpleasant then 2) offer something mostly pleasant but at small cost to yourself. After hearing about the breadsticks being out and the refund offer, I was preparing to counter-offer getting some chips or something more substantive. However, since he quickly offered me a better alternative, I figured it was fine.

The woman as third party was an interesting element. I am typically insusceptible to guilt, although something about her mannerisms made me feel afterwards that she somehow influenced me to accept the offer without offering my own. The fact that I have a small established relationship with this business means I feel pretty good about the whole situation though.

Could Subvocalization Lead to Readability?

A common writing tip is to read your writing out loud. I thought this was corny, but tried it out. When reading my work aloud, it was easy to spot where the sentences were too long or the flow was not quite right. I was actually surprised at how much better my writing sounded after doing this phase of the editing process.

My main idea or thought is: could subvocalization by writers lead to increased readability?

Subvocalization is explained well on Wikipedia. I generally try to avoid subvocalization when reading because it allows me to read faster. However, it could be that writers–who are generally quite competent readers–would benefit from more subvocalization to hear their words. Indeed, if a large portion of their readers subvocalize, then they have more incentive to fit the medium to the audience by making the words match the internal ear better.

Overall, I think the separate stage of reading my words aloud works well for me. It was just an interesting train of thought that I wanted to share.

Systems Designed to Discourage Engagement

@jchyip: “There is no apathy, only systems designed to discourage engagement”

I think this is an insightful quote. I would quibble with the idea that all systems that discourage engagement are “designed” to do this. I would rephrase to “There is no apathy, only systems whose design discourages engagement.” I think it implies too much conscious manipulation to create a system that discourages engagement. Kind of related to “do not attribute to malice that which can be explained by ignorance” idea.

However, it’s certainly interesting to consider whether the current political environment or a workplace was deliberately manipulated to discourage engagement or it was just an emergent, evolutionary, or incremental effect. Or how a discouraging system can be tweaked to be more hospitable to engagement. Further, the quote seems relevant to creating company culture, in that we want to create a system that rewards and encourages engagement. Starting with the right principles seems to be better than winging it and then changing over time, as the system reinforces what is already there.

Categories: main

« Why You Should Explain Your Android App's Requested Permissions Systematizing Dinner »