More on Moore’s Law

If you’re like me, you probably have a bunch of addons for Firefox. I have between 5 and 20, depending on whether I’m running the latest beta (3.1 beta 3), or the stable version.

The problem with this, as I have just experienced, is that you not only increase the chance of failure, but you also increase the cost of repair. Let me explain.

Something happened to my Firefox installation, and anything typed in the location bar would grind Firefox to a halt, then eventually crash it. I tried re-installing it, but the problem remained. I then upgraded to Firefox 3.1 beta, without any further success. Since Firefox has such a large install base, I assumed someone else would have the same problem, and eventually there would be a fix. The plan was to use Safari (4 beta) until then.

However 3.0.9 came out and when I installed it, Firefox didn’t work any better. I resolved myself to do some sleuthing, and eventually figured the problem out, and solved it. The issue was caused by a mix of Google Gears, Adblock, PasswordMaker, and some others.

What’s to learn from this?

Customization is costly. Maintenance costs and instability are exponential to level of customization: if each component has a 1% chance to break per month, and you have 20 components, then you have an 18% chance of failure per month (1 – .99^20). Bear that in mind when you customize anything.

You also lose in security: the more you customize your software, the more lengthy and intricate it becomes to upgrade to the next version. This means that you run software that is always a few patches behind, and you become an easy target for hackers. The vulnerabilities of the version you use are public and easy to exploit.

Rising Probability of Similar Error

Furthermore, the more you customize, the more you reduce the userbase of similar installations. If you use default Firefox on default Ubuntu, then whatever error you encounter, millions of others will as well, and you can count on a patch to soon follow. If you use a highly modified version of Firefox on Ark Linux, you’ll have to do the detective work and patching yourself.

A corollary to this is that it makes an excellent argument for buying Dell when considering using Linux. You’ll have much less problems as the userbase is much larger. If you use a Fujitsu LifeBook 7010, with a tiny userbase, you’ll have many driver issues.

Analogy to Firefox addons

Coming back to my Firefox addon example, I could have waited years until this problem was found and solved. I would have to wait until someone else using Firefox AND Gears AND Adblock AND PasswordMaker AND Mac OS X find the error, come up with a fix, and share with all. This means that you just can’t wait bugs out if you aren’t in the mainstream. Whereas you can if you use plain vanilla Firefox on Ubuntu.

EDIT: Apparently, this is Oracle’s approach. Paraphrasing them, they advise to “stick with us; we will do it all, and make sure everything works together, and do not customize as you can not then take advantage of our innovations“. This goes in the same direction as what I argue in this post. Customization is costlier than it appears.

Some time ago, a friend suggested I enter the YouOnTheWeb competition organized by Revevol.eu. All I needed to do was give my name, phone number, and email address to gain entry. So I did. The Jury would then google for all participants, and award the one with the best online presence with a Samsung NC10 Netbook.

I was told Monday that I had won, much to my surprise.

I am to meet the Revevol team on Friday next time I am in Paris, and there will be some picture-taking and formalities. I look forward to it, as Revevol is doing something similar to what I am doing: promoting the use of the Cloud in the enterprise. They provide services in adopting Google Apps and other SaaS vendors. I, with Cloud in Code, provide services in adopting Cloud Computing.

So I went home with my brand new Samsung NC10 netbook, wondering what I was going to do with it. It had Windows XP pre-installed. I thought to myself that I would keep it that way, since all I need is Firefox. 4 restarts later, I started downloading Ubuntu Netbook Remix out of frustration. It is amazing how three service packs later, Windows XP is still crap. However the Ubuntu page dedicated to the NC10 mentions some functionality lacking until the next kernel, so I’ll wait for 9.04 due this month (Jaunty Jackalope).

Incidentally, Mozilla Labs announced the general beta availability of Weave, a Firefox addon that lets you sync your Firefox profile across multiple computers (only available on Firefox 3.1 beta). This was a perfect occasion to install on both my MacBook Pro and the NC10. And it works, allowing me to get the same results on the Smart Location bar, sync open tabs, keep same bookmarks and passwords (I use a unique password for every site, a hash of the site url concatenated with a master password, a pain on mobile devices). Works great, highly recommended if you have multiple computers.

It can be argued that without signaling, Akerlof’s Market for Lemons would vanish, or at least all innovation and delta quality would be driven out. Akerlof, economics Nobel laureate in 2001, described how:

the interaction between quality heterogeneity and asymmetric information can lead to the disappearance of a market where guarantees are indefinite. In this model, as quality is undistinguishable beforehand by the buyer (due to the asymmetry of information), incentives exist for the seller to pass off low-quality goods as higher-quality ones. The buyer, however, takes this incentive into consideration, and takes the quality of the goods to be uncertain. Only the average quality of the goods will be considered, which in turn will have the side effect that goods that are above average in terms of quality will be driven out of the market. This mechanism is repeated until a no-trade equilibrium is reached.

Why is it that there is a market for lemons, like in used cars, when there are reasons why there shouldn’t? Signaling provides some light to the theory.

In Spence’s winning paper Signaling in Retrospect and the Informational Structure of Markets, education is seen as a signal in the labor market. From his Prize lecture:

The idea behind the job market signaling model is that is there are attributes of potential employees that the employer cannot observe and that affect the individual’s subsequent productivity and hence value to the employer on the job.

He then introduces the basis for the mathematical model subsequently built:

Let us suppose that there are just two groups of people. Group 1 has productivity or value to any employer of 1, and Group 2 has productivity of 2. In this example, these productivity values do not depend on the level of investment in the signal. If there were no way to distinguish between people in these two groups then if both groups stay in the market, the average wage would be 2 – α, where α is the fraction of the population in group 1, and everyone would get that wage. If the higher productivity group through dissatisfaction or for any other reason, exits this labor market, the average productivity and the wage drop to 1. This phenomenon when it occurs is sometimes called the adverse selection problem, a label most commonly applied to insurance markets. It is structurally the same problem that Akerlof described in his famous paper on used cars (lemons).

Why is Education an Honest Signal in this famous Nobel winning paper?

Quite simply, because it is expensive and as such not easily falsifiable.

The job-market model is a way of looking at product pricing and markets. The analogy is clear: Job seekers = Complex product offering, Skill level = Product Utility, and Wages = Price, but Education = What?

Education = Marketing!

What is the purpose and value of Marketing? Marketing is sometimes treated as a necessary evil, something we would all rather not have to truck with and most certainly not something that adds “real” value to the product. Wrong! The value of Marketing seen as attempts of Honest Signaling is huge:

• Branding. Branding is a form of honest signaling. It is a promise that is hard to break. IBM cannot afford to put its name on a dysfunctional product, as this hurts it. You can therefore trust software IBM-branded software.
• Advertising. Advertising has lost most of its credibility, and has equally lost in persuasiveness. Despite this, it can be seen as a mechanism where the signal itself is the value, not the content. The “It must be good since they can afford to blow so much money on advertising this stuff” concept, first introduced by Nelson (1974) who described the idea of advertising signaling quality by dissipating part of the profits, then formalized by Kihlstrom and Riordan (1984), and Milgrom and Roberts (1986). Advertising is a form of honest signaling.
• Analysts. The industry of third party analysts, man in the middle, and honest brokers à la Gartner. More on this later.
• Without signaling, there would be no point in developing better products, as they would not command higher prices. The only rational decision would be to cost reduce.

Marketing -now defined as the signaling body- has developed new signals over time: adverts, public relations, and now social media. All of these are tactics of influence, but more usefully and insightfully defined as attempts at Honest Signaling. This creates a new framework from which we can find better ways to send Honest Signals to the market, and thereby capture this huge value.

Spence’s Job-market model provides an example for opaque products, and how they require a signaling mechanism to provide information. Opaque products require honest signaling, and so need to be marketed with that in mind. This means using analysts, branding, etc. is not a waste of resources. It makes rational sense for players to hire an honest signaler.

So, what is the value of Honest Signaling?

Honest Signaling allows the marketer to provide information on a product’s quality in a trustable manner, where it would otherwise be costly or difficult to do.

For example, how do you demonstrate that your data integration software is good?

• You cannot educate the market. Evaluating such products takes weeks, if not months. Customers do not have time to do thorough research on competing products.
• You cannot advertise. As mentionned before, advertising in its traditional form has lost its credibility, and therefore would not provide significant advantage.

So what options do you have?

• You can put a big brand name on it. A corollary to this is that it can make sense for a large company to purchase a small one in order to re-brand its product. In this specific case, the purchase might in itself be a sufficient Honest Signal, as it would not purchase it if it weren’t good.
• You can hire an analyst firm. The analyst firm then evaluates your product, and you can purchase the evaluation for display. If your software were poor, you couldn’t afford to do so as you would get a negative and detrimental evaluation. And so this is an Honest Signal.
• You can expose yourself to risk. Best example of this is a guarantee. Make an expensive promise and uphold it. If your software were poor, this would be prohibitively costly.
• You can fill the airwaves with advertising. Show that you can afford to squander resources. The key is not content, but volume.

I was visiting my inlaws the other day, and the alarm sticker on their front window caught my attention. It is one similar to the one depicted here.

It got me thinking about the effectiveness of such a measure. How likely is a thief to believe it? If you were one, would you?

For the sake of this analysis, lets assume that we are dealing with a rational thief. This excludes thieves under the influence, junkies, and a couple other categories. The rational thief is confronted with a choice of trusting or calling a bluff when he sees the sticker. If there is no means to verify the presence of a working security system, then he must rely on the signal alone (if he can verify that there is a security system, then he should move on to an easier target).

The Theory

Out of the total population of people displaying the sticker, if A is the number of people with a working security system, and B the number of people who don’t, then the probability that there is one when randomly picking a house is α = A / (A+B). If we furthermore assume that a house with a security system means a certain outcome of being caught by the police and jailed (we will analyze the effect of uncertain outcome in a later post), then the thief can determine his optimum strategy.

Strategies and equilibrium

If α = 0, the thief will never be caught as there is no security system. His strategy should be to always pick the house with a sticker. If α = 1, then he will always be caught. His strategy should be to never pick a house with a sticker.

But somewhere in between, there is an equilibrium. This equilibrium depends on the payoff, of course. To the extreme, you have a payoff so high that it is worth going to jail for, which results in the outcome strategy of always disregard sticker. On the other end, if there is no payoff, then the outcome strategy is never disregard sticker. Furthermore, the payoff is always relative to the thief ($100 worth of stolen goods does not have the same appeal to all thieves), especially true when considering a professional thief and an occasional one.

Sometimes doubt is enough

You could argue that sometimes, introducing doubt into a rational analysis is enough so that the thief prefers to consider an easier prey – a house with no sticker for instance. This is the rationale behind Microsoft’s famous FUD (Fear, Uncertainty, and Doubt) tactics. People will stick to actions of which outcome they can predict rather than those they cannot. Microsoft’s tactic has been the introduction of uncertainty in the outcome of switching operating system from Windows to Linux. Experience shows us that uncertainty does not need to be grounded, only perceived. The company introduced stories of failure and that proved sufficient.

When it comes to our analysis, displaying a security system sticker produces doubt in the thief’s mind. This most likely increases the likelihood of him moving on to an easier target. If you and a friend are chased by a lion, you do not need to outrun the lion: you only need to outrun your friend. Same goes with your neighbor’s. You only need the thief to perceive that your house is better secured than his. This falls outside of the Handicap Principle and consequently Signal Oriented Marketing, however.

Seem familiar?

If you’ve been following this blog, or know a bit about the Handicap Principle / Signal Oriented Marketing, then this analysis probably seems familiar. It involves mutually distrustful parties (the thief and the house owner), that communicate through a signal (the security system sticker).

The signal between the two parties conveys information about the likelihood of being caught by the police, similar to the gazelle’s stotting about the likelihood of the chasing predator not catching it.

How to bring credibility

Through verifiability

In some specific cases, the presence of a security system can easily be made verifiable. In those cases, it should, similar what Stanford suggests webmasters do in article 1 on its website credibility guidelines. Thus a solution would be to make the security system (cameras) visible.

But in other cases, such verifiability is not as easily attained. If it is not possible for a thief to verify the presence of a security system and he must refer solely to the signal (the sticker), than a system in which each sticker has a unique identifier issued by the security system vendor that, when anonymously given by SMS or Internet to the vendor, would retrieve the house address and validity of the sticker.

The combination of address and validity achieves two things. First, it prevents duplication of stickers, and usage on more than one location. Second, it prevents people without a security system from faking the signal. It ties each reference to a unique location, and makes the signal perfectly trustable.

However this system is complex and relies on the thief’s knowledge of the system, as well as him trusting the anonymity of the lookup process, both of which are not a given. In such cases, it is preferable to use Honest Signaling.

When not verifiable

Honest Signaling allows credibility to be established where the quality of an object is hard to ascertain. If it is not possible to verify some quality of an individual or organization, then there must be something about it (the signal) that conveys information on the quality, in a trustable manner. The way nature has solved this, is by creating handicaps. Signals that are costly in relation to the interaction of parties. This means that there is a penalty if the signal isn’t trusted. You show your strength by imposing a handicap on yourself. Like handicaps in Golf.

If you buy into the concept of biomemetics, this translates to a simple tactic. The signal should welcome thieves to “test our security system here”. Or provide directions towards the security cameras. It should give up information that would be useful to the potential thief. It should impose a handicap upon itself.

You may find this behavior in people being boastful: the security guard so confident that the vault is secure, that he openly talks about secret details on the motion detectors, etc. The opposite is telling as well. If he were unsure about the security of the vault, he wouldn’t talk about it at all.

Conclusion

If you (the reader) were to remember one thing about this rather lengthy blog post, it is that Honest Signals are useful a) when you must convey information on the quality of something to someone who does not trust you, and b) when the quality of that something is hard to measure or ascertain. In that specific case, then you should impose a handicap upon yourself, that is costly in relation to the outcome of the interchange.

мебели пловдив

Lets say you’re working on a software project; something new, some never-seen-before app that will change the way things are done.

At some point, you’re going to ask yourself, what features should I add? What should I be working on? You’re going to be determining your priorities.
The common approach is to ask your clients or userbase what they would like to see in the next version. There are many good sites that do this (uservoice.com, ideascale.com, and crowdsound.com are a few), and you’d basically be crowd-sourcing your product management / feature development.

In an enterprise setting, the process is somewhat similar. The sales guy call his lead up and does his sales technique. If the sale goes through, fine, but if not, he tells his boss. His boss then asks him why he didn’t sell, and he turns around and asks the lead why he didn’t land the sale. Lead then responds something along the lines of “your product didn’t have feature X”. Boss then makes a report, and sends it to engineering, which then implements feature X.

This method has a clear fault: leads can and will respond with all sorts of excuses to end the conversation. You then end up building your product on whatever your leads come up with to stop talking to your sales team, and this is how bloatware is born.

There needs to be a counter-force to suggesting a feature. A cost of sorts.

If you start out solving your own problem, then all is well. You only do the essential stuff, what counts to you, because you bear the cost yourself. But after a while, your own problems are fixed, and you move on to fix other people’s problems.

The cost can be anything, but energy expenditure is a good way to determine whether there is cost at all. This energy expenditure can be time. Development resources (sending in a patch), or time working with you on a solution. Or it can be financial. From $10 (or even $1) for individuals, to thousands of dollars for companies.

Since there is a drawback to suggesting a bugfix or creating a new feature, you avoid entirely the stuff nobody wants. And the stuff you really need, you’ll find someone willing to shell out the $10 for the feature. If you have a freemium (free + premium) business model, you may want to suggest going to a higher pricing tier in exchange for development of said feature.

For those who want to know about the theoretical underpinning of this, it is part of my Master’s Thesis research on the Handicap Principle from evolutionary biology applied to Signal Oriented Marketing. It studies how trust emerges among mutually distrustful individuals through specific types of signals. These signals that establish trust are Honest Signals, in the form of handicaps. Like why you would buy a $10 hammer with a lifetime warranty, even though it will never be worth your time sending in a complaint for a refund. Or why witches have long noses, and clowns have round ones. Or why we wear ties at formal gatherings, but bowties in state dinners.

I went to a negotiation competition last Thursday called Les Négociales. We were 120 business school students from the Northern France region.

What was interesting wasn’t so much the competition itself, as the sight of recruiters and head hunters flocking to competition finalists. Competition achievement prowess being an Honest Signal for proficiency at work.

Michael Spence wrote in his Nobel Prize paper ‘Signaling in Retrospect and the Informational Structure of Markets’ on the influence of Education in the Labor Markets. Seen through the lens of Honest Signaling and the Handicap Principle, it would appear that in reality, Education is a Handicap. A handicap that is an honest signal of ability to perform physically and / or intellectually. Or a signal of productive capacity. Likewise, Competitions are a handicap. Both Education and Competitions consume an individual’s resources and produce in themselves no value.

It is interesting to see something everybody knows in a different light.

We have seen in a previous post What the handicap principle is. However an issue arose on the exact meaning of Cost of a Handicap.

A handicap must be Hard to fake, not directly Costly. Cost is only in relation to the interchange.

What is a cheater?

In Handicap Principle jargon, a cheater is an individual that emits the signal despite being unfit. This faking must be difficult to achieve for an equilibrium to be obtained between fakers and honest signalers. Faking is hard and a whole school around self-deception (like Trivers) is based on this.  Deceiving yourself makes it easier to do.

For example, a handicap is only costly when you fake it badly and you get caught. In nature, this is when the Lion chases the Gazelle despite the signal, or when an employer asks for the diplomas you added on your résumé but do not actually have. There is no cost in emitting the signal, only in your bluff being called on.

In Spence’s paper ‘Signaling in Retrospect and the Informational Structure of Markets’, the signal that is education has a cost though. The trait that is signalled is productive capacity, and one signal for it is education. More on this in the post Honest Signaling in Labor Markets.

If you’ve ever dealt with the advertising industry, we probably know about the cognitive threshold under which you do not gain much by advertising, and over which you enter the prospect’s mind. This threshold determines whether the prospect will be able to retain and recall anything from the information advertised.

The handicap principle shines a new light on our understanding of this phenomenon. The threshold can be understood as the minimal expenditure required to become a handicap, and honest signal of quality. The sheer amount of advertising you do. The message is not the content, but the amount. A sort of “if they can afford to spend so much money on advertising, it must be good” school of thought.

According to professor Gilbert in his Lottery winners and Dogs & Pigs on a leash papers, we go back in out memory and try to recall the frequency of images that pertain to what we want to assess. This is why you need to concentrate the message in time, and why a threshold is needed in a marketing campaign.

Model for a Cheat

A corollary to this is that there is a model for a cheat.

In the above advertising example, the cost of advertising is born by the enterprise advertising. In nature, antlers have to be regrown each mating season, gazelles cannot borrow energy from the future. There is a distinct time slot for each signal. No spill over allowed. This is the frequency (amount over time) of advertising, or antler size in a season.

You can break this model if you break this natural time slot. Ponzi schemes like Madoff’s did precisely this. They signalled success by stealing from the future. The reason us humans are vulnerable to this, and so many people get caught in similar Pyramid schemes, is probably that our brains aren’t geared to detect this effectively.

Another example can be taken from my childhood experience in my family’s summerhouse north of Copenhagen, in Denmark. Most of the houses in the neighborhood are heated with an oil furnace. You have an underground tank outside that is filled up on a regular basis. I remember one company that did this service changed its model from filling up when empty, to everyone in the same area on same day. This was very clever as people saw the oil trucks everywhere that day. They concluded that the company had to be doing well and be a good company since there was so many. They forgot they didn’t see any next 2-3 month.

In this case, the time slot is broken. The distribution over time of occurrences is uneven. The cheat is to steal from future occurrences to create high frequency point events.

So bear this in mind both when receiving and emitting honest signals. Ask yourself, where does the cost of the signal come from?

EDIT: I read Dominique Olié Lauga’s PhD thesis titled Essays in Behavioral Industrial Organization, Corruption, and Marketing, in which she argues that advertising has an intrinsic value. She comments that “this is in contrast with the view that advertising is a pure money-burning device”, referring to Nelson (1974) describing the idea of advertising signaling quality by dissipating part of the profits, that Kihlstrom and Riordan (1984) and Milgrom and Roberts (1986) formalize. This is similar to Nobel laureate Michael Spence, who first describes a model in which education does not affect productivity, then adds intrinsic value to it in a more elaborate model.

In this entry, I’ll share with you how to detect in a person what drives them.

The head of our marketing major at ESC Lille, Christophe Sempels, has brought us exceptional speakers: the first one was Arnaud Pêtre on neuromarketing, in November last year. Today, we received Pierre Moorkens on Developmental Psychology. He talked to us about humans, what drives them, and how to understand the person we have in front us. Be it spouse, friend, or colleague.

First, you must understand the distinction science makes between Character and Temperament. Temperament is acquired during the first 3 months after birth, Character is acquired onwards until between 18 and 22 years.

Temperament drives people through the pleasure they take in doing something. Character drives someone only so long as they succeed in the undertaking.

Take your relation to order. Say you just finished cooking, and your kitchen is messy. Do you clean up because you like to? Or because you don’t like the mess? Or do you think to yourself that you should clean up, but now’s not the time?

In the first case, you clean up because you like to. It’s in your temperament. In the second, you do it through aversion. You were taught or influenced in disliking messes by your relatives. In the third, you postpone cleaning up, because you don’t really care. You know it needs to be done, but there is nothing compelling to do it. If you do it because you like it, it is temperament; if you do it because you have to, it’s character; if you aren’t compelled, it’s neither.

So what?

Well, lets say you’re hiring an accountant. You ask him about what he likes in accountancy: he answers he likes precision, numbers, calculus, etc. You then put him in an imaginary scenario where he takes over from an accountant that made a poor job. His reaction to this situation will determine how he’ll fare.

If he mentions disliking numbers that don’t add up or anything similar – stay away from a hire. You are looking for someone who will see this as an opportunity to make it right.

Say your company acquired another one. If the acquired company’s books aren’t perfect or don’t match the formatting of your own, your new accountant will be stressed. And will perform poorly.

If you do something aligned with your temperament, you enjoy the process. If you do something aligned with your character, you enjoy the results. You want to hire someone who enjoys the process.

Doing something aligned with your character consumes energy, doing something aligned with your temperament produces energy. If you consume too much energy, you burn out. You then need to fall back on something that produces energy to restore morale.

So what this means is that when you are hiring someone like a software developer, try (among the many other things you need to do) to figure out his temperament. See if there is a match with the job you are offering. He enjoys making things run fast? Give him a job optimizing code. He likes making new things? Assign him to a new product development projects. This might seem trivial, but this is often overlooked.

As you can imagine, a good indicator of motivation are the person’s personal projects: the ones he does for pleasure.

Disclaimer: I did some research, and there does not seem to be any consensus on this. And I have never yet hired anyone. It rang true to my ears, that’s it. So take with a grain of salt.

In this entry, I’ll share with you the tips Pierre Moorkens also gave us on how to pull someone out from stress.

Stress is a transfer of information processing from the pre-frontal and limbic regions of the brain to the older, primitive reptilian brain. This region processes life-or-death situations, where you must Flee or Fight. There is also another form of stress, in which you inhibit yourself and seek assistance (Abandon).

Fleeing individual – recognized by his scratching himself, saying yes to everything, wanting to end the conversation. You want to position yourself as an accomplice.

Fighting individual – recognized by his eyes, fixed on you. Anger, confrontational. You want to acknowledge his grief.

Your colleague Jeff is angered by his coworker: his know-it-all attitude, his laid back attitude, everything. It’s impossible to work with the coworker. Your best shot at getting Jeff back on track is to agree with him: the acknowledge the coworker is laid back, not easy to work with, and self important. That should get Jeff calmed down, and you can take it from there.

Inhibiting individual – recognized by his eyes, looking downwards. You want to comfort him.

Eric has been missing deadlines, so you ask him what’s up. He looks down and says not much, work as usual. You talk to him about the deadlines, and he says that you should ask someone else to do the job, he isn’t fit for it. Jeff would be better. Project is too difficult, too complex. In this case, you should agree with him that the project is indeed hard, tricky, and complex. You should comfort him until you pull him out of the stress.

Disclaimer: I did some research and there does not seem to be any consensus on this. Take with a grain of salt.