Regular readers may remember that I posted on YouTube a little video of some caterpillars in my front yard a few months ago. I went back there a few days ago and took a look at how many times it had been viewed.
OVER 160,000.
This is obviously a much bigger audience than I will get for anything else I do in my entire life. I’m sure there is a moral in here somewhere, but I have no idea what it is.
Here is a common-sense approach to software development, which I’ll call the checklist approach:
This essay uses shows why the checklist approach fails: why adding useful new features to a software product can make the overall product less useful to end users. This is a phenomenon I like to call featuritis.
At its simplest, a software product is a set of n independent features (i = 1,…, n). Each feature has a utility to the customer of ui. The overall utility of the software product to the customer is therefore
U = Sum( ui )
This tells you to keep adding features to the software to increase its utility, end of story. Not very interesting.
But software is not quite so simple as this. Each feature has not only a utility, but also a cost — ci. This is the cost to the end user—not the developer—of using or deciding not to use the feature. The cost may include the time taken to find out about a feature, deciding how to use it, whether it is the right feature to use (among the options available), difficulty exploring and evaluating the feature, and so on.
With this cost added in, the overall utility of the software product to the end user is
U = Sum( ui – ci )
Let’s keep things really simple, and assume that all features have the same utility (u). We can’t do this for the cost though: the cost of using a feature inevitably depends on the total number of features in the product. This may be because it takes longer to find information about a particular feature in the documentation or user interface, or because it is more difficult to decide if this is the right feature to use among those that are present, or a host of other reasons (more on this below). So the cost of finding out about a feature is (c * n).
The overall utility of the software is then
U = Sum( u – c * n )
or, as all the terms are equal,
U = n ( u – c * n )
Which looks like this:
The utility function increases up to a maximum at n = ( u / 2 c ), and then decreases: that is, beyond a certain point, adding new features actually makes the software less useful. In other words, software is vulnerable to featuritis: it can become so complex that its complexity makes it useless. It may contain useful features, but finding these needles in the haystack of the product is frustratingly difficult. Experience tells us that this is a reasonable, if not surprising, result.
What is more, adding features that are useful in and of themselves can still introduce featuritis. For a feature to be useful, its utility (u) must be greater than the cost of finding out about it (c * n). So using this model, features will be added to the software until the following condition is met
n = u / c.
which is where the line crosses the x axis. The maximum utility of the software occurs at half this value (n = u / 2 c): if we continue to add features until the last feature is only just useful, the overall utility of the software will be U = 0. If features were stopped at (u / 2 c), the utility would be U = ( u2 / 4 c ). Even a useful feature degrades the usability of other product features, by making them harder to use (increasing the cost of using them).
It follows that checklist driven software development will lead to poor software. Checklists are simply lists of useful features, without any consideration of the costs they introduce to the customer. A longer checklist is often assumed to be intrinsically better than a short checklist, but we have just seen that this may not be so.
What it is about this very simple model that produces these results? The key assumption is that features have independent utilities, but that the costs of using features are not independent. In economists’ jargon, features exert a negative externality on each other. Is there a reason to think this might be true?
The independence of feature utility is simply a matter of defining a feature
properly. For example, if a very common scenario requires two "features" (A and B) then the utility of feature A depends very much on whether feature B is present or not. By itself, the utility of feature A may be very low — you can’t do much with it by itself. Once feature B is present, though, feature A becomes very useful. They are not independent features. The problem here is that "features" A and B are incomplete, and so are really part of a single properly-defined "feature". The model requires that we define features in terms of tasks that customers can carry out. The model does not tell us to implement 5/6 of a feature and then not implement the last sixth because of complexity worries.
The second assumption is that the cost of using a feature is dependent on other product features. The idea of a cost of using a feature is a very general one, and is not only "how long does it take to locate this feature in the documentation?". It may also reflect the confusion and uncertainty introduced by a plethora of choices. For example, if there are multiple ways of carrying out a task, the customer must decide which way is the best. If there are other features that appear to be related, the customer must investigate those (and discard them) before deciding on a course of action. If there are prerequisite features that must be understood, the customer must learn these also. They must spend time evaluating the alternatives. It seems reasonable to assume that the cost of wisely using an appropriate feature does depend on the overall complexity (number of features) in the product. The job of user interface designers and documentation teams is, at least in part, to minimize this destructive interference between product features. Good UI and documentation can help postpone the point at which featuritis sets in, but can’t hold it back for ever.
Of course, while simplified models like this might help us watch out for certain kinds of trap, they can’t help us decide which specific features to include, and which to discard. Also, there is not much to be gained from trying to pinpoint the particular u and c associated with features or products. Despite the equations, it is a qualitative model and cannot be easily quantified in a useful manner. Finally, while it is certainly true that ui and ci are far from constant in any product, there is probably not a lot to be gained by trying to refine their representation. As soon as models like this are made more complex, the result is a very open-ended and conditional prediction. Building fine software products can’t be reduced to equations.
But I do think the central ideas are broadly correct: complexity does influences cost and utility in different ways, software does tend to become overly complex. and—most importantly—asking "is this a useful feature?" is not the right way to develop good products.
Yappa Ding Ding wrote a lively post about automating intellectual jobs in response to my gripes about lawyerbots. Yappa is more optimistic than I was being about the impact of this automation. As she says:
Wouldn’t it be cool if we all had our own lawyerbot to protect our
interests and automatically communicate with other lawyerbots. Just put
my libel case winnings in my bank account, please!
…Similarly, we can and probably will automate vast chunks of what is
done by lawyers, doctors, politicians, bureaucrats, engineers, computer
scientists, and so on. This could lead to a reduction in prices, just
as manufactured goods are much cheaper than they used to be. That could
be important. For example, now, if you are charged with a crime and
have enough money that you are ineligible for government-paid legal
assistance, you will likely go broke defending yourself. Ditto if you
get involved in a contested divorce settlement. Even handling a real
estate transaction costs hundreds or thousands of dollars in legal
fees, when most of the work is rote. It would be a social revolution if
the cost of getting legal advice became more reasonable.
So was I being unnecessarily grumpy or just necessarily grumpy?
Well, probably yes a bit of both. The picture that Yappa paints is definitely cool. Right now justice is, as someone once said, open to everyone in the same way as the Ritz Hotel, so making legal help cheaper is going to help a lot of people. For example, JS showed me this thing called Eulalyzer which "reads" those pesky end-user licence agreements and, well, let’s let it speak for itself…
EULAlyzer can analyze license agreements in seconds, and provide a
detailed listing of potentially interesting words and phrases. Discover
if the software you’re about to install displays pop-up ads, transmits
personally identifiable information, uses unique identifiers to track
you, or much much more.
This is a Good Thing, and I’m sure there are lots of other ways that software could lower efforts and costs, especially routine documentation checks, like real estate transactions and so on. Why pay lawyer rates for a person to read through these routine documents when software could go through them and flag any unusual text. Great.
But the down side remains. There are two parts to most legal exchanges – one side prepares a document and delivers it; the other side reads it and responds. While low cost makes it easier to do the reading and responding, it makes the preparation easier as well. And more than anything, it makes the delivery easy – and this is where the problem really lies: there is a prospect of spamlike legal threats, warnings, and so on because they can take advantage of the cheap production of documents and then multiply that cheapness (if that’s the right phrase) to scatter them widely.
What we really want, of course, is the one without the other. E-mail without the spam; downloads without viruses, networks without trojans. Not a lot of chance of that, but perhaps we can minimize the downside of some of the new developments if decide that spam legal threats are invalid.
Kieran Healy, Last Best Gifts, University of Chicago Press, 2006.
Gary Becker and Julio Elias, Introducing Incentives in the Market for Live and Cadaveric Organ Donation, (working paper).
The
distinguished theoretical chemist John Murrell wrote that physicists like to solve simple models exactly, while
chemists like to solve detailed models approximately. There are
benefits to both approaches. Take the study of solid state electronic
and magnetic properties, where the early work of physicists was to
compute, using very sophisticated techniques, the properties of highly
simplified models such as a free-electron gas with a uniform background
of positive charge. Chemists, on the other hand, were busy studying
complex materials with intricate structures, but didn’t have the theory
to do more than classify the kind of observations they made. As a
result, physicists gave theories for superconductivity and other exotic
phenomena, but because their theories had so little information about
the specifics of molecular structure they couldn’t predict the kind of
material where superconductivity would be found. The discovery of
high-temperature superconductors in the 1980’s had physicists as well
as chemists scratching their heads. The empiricism of chemistry (where
theorists form about 10% of the population) compared to physics (where
the number is, I think, more like 30%) forces theoretical chemists to
deal with the messy and specific problems that their discipline’s vast
body of observation and experiment demands.
Economists are to
sociologists as physicists are to chemists. Whenever you see a
discussion of scientific method and rigour in the social sciences, the
comparison is always made to physics. This is not surprising, because
physics remains the archetypal science, but it’s a bad thing because
the social sciences — like chemistry — have to stay grounded in the
empirical side of what they study. The details of any particular
problem are so often crucial to the outcome that you can’t use "the
light touch of the physicist" and expect to come out with predictions.
It’s all very well to talk in the abstract of markets, but when reality
hits it’s the details that often matter. Sociologists spend their lives
looking at those details, like chemists they tease out conclusions from
the structure and dynamics of particular circumstances and events. And
like chemists, they get less respect than their more formal, more
model-driven, and less empirical siblings.
These different
proclivities are on show when sociologist Kieran Healy looks at blood
and organ donation in his new book Last Best Gifts and when economists
Gary Becker and Julio Elias look at organ donations in their widely
discussed working paper, and my ex-chemist self is glad to say that
Healy comes out looking better.
The basic problem is simple.
Advances in surgery and in immuno-suppressing drugs have made organ
transplants safer and cheaper since the 1980’s. The number of organ
donors (whether live or dead) has grown, but not nearly enough to keep
up with the demand, and the result is a growing shortage of organs.
People who could be saved are dying while waiting for "donor" organs.
What can be done?
Becker takes, as anyone who has read any of
his writings would expect, a forthright and straightforward approach.
Shortages mean that supply and demand don’t match. As he says in a blog essay on this issue:
To an economist, the major reason for the imbalance between demand
and supply of organs is that the United States and practically all
other countries forbid the purchase and sale of organs. This means that
under present laws, people give their organs to be used after they die,
or with kidneys and livers also while they are alive, only out of
altruism and similar motives. In fact, practically all transplants of
kidneys and livers with live donors are from one family member to
another member. With live liver transplants, only a portion of the
liver of a donor is use, and this grows over time in the donee, while
the remaining portion regenerates over time in the donor.If laws were changed so that organs could be purchased and sold,
some people would give not out of altruism, but for the financial gain.
The result would be an increased supply of organs. In a free market,
the prices of organs for transplants would settle at the levels that
would eliminate the excess demand for each type of organ.
You
make supply match demand by introducing a market — paying people for
their organs (eg, kidneys) or for the organs of their family members
after death — because that’s what markets do. So he and Elias do some
rough calculations on what would be needed ($32,000 per liver, while
the current cost of a liver transplant is $175,000), and say "let’s do
it". Becker is under no illusion that this proposal will be adopted
soon, but believes that there is a real chance that it may be taken
increasingly seriously over the coming years, and serious discussions
are increasingly (so I’m told) including markets as possibilities, as
in a recent issue of Kidney International.
There
are no details in Becker and Elias’s sketch of the form that the market
would take. Of course, Becker is aware of many of the arguments against markets
(many coming from Richard Titmuss’s influential 1971 book The Gift Relationship)
and returns to them in his later blog posting to address them: that
"commodification" of body parts is immoral; that payment may drive out
altruistic donations and so not yield the bumper crop he predicts; that
organs may be removed mainly from the poor and installed mainly in the
rich; that organs may be removed forcefully from people (as Falun Gong
supporters are claiming is happening in China now) in order to be sold;
that people may regret an impulsive and irreversible decision to sell;
that payment may lead to people lying about their medical health and so
lead to infected organs entering the system. He doesn’t so much argue
these issues as dismiss them. Tellingly, he uses the passive voice: the
quality of blood "can be maintained at a high level", the source of
organs "could be determined in most cases without great difficulty";
the number of impulsive donors "could be sharply reduced by having a
month or longer cooling off waiting period". It isn’t quite clear who
has the incentive to maintain all these standards, or carry out these checks, or how
much it would cost to persuade someone to do so. And yet in markets for experience goods
(and organs, surely, are experience goods of a visceral kind)
information issues are at the heart of the problem. The cavalier
brushing aside of issues of trust, fair dealing, and asymmetric
information makes Becker’s case unconvincing to this reader.
Becker also commits what Tyler Cowen of Marginal Revolution calls the libertarian vice:
assuming that the quality of government is fixed. In this case, that
means assuming that if altruism isn’t working now, then it won’t work
in the future. "If altruism were sufficiently powerful, the supply of
organs would be large enough to satisfy demand, and there would be no
need to change the present system. But this is not the case…"
Kieran Healy spends most of Last Best Gifts
exploring the very things that Becker skates over so casually, and argues
that they are the heart of the matter. Altruism is not a fixed
quantity, but depends crucially on "the cultural contexts and
organizational mechanisms that provide people with reasons and
opportunities to give" (p2). Those unspecified actors who are the passive
voices of Becker’s arguments are, Healy argues, the key to success or
failure when it comes to blood and organ donation. For example, organ
donation from the newly dead is, in practice, dependent on approval
from the relatives and the rate of approval depends in turn, it turns
out, on who asks them. If the person who is helping them come to turns
with the death is the one who asks them to approve donation of the
organs, they are more likely to refuse than if somebody else (even from the
same organization) asks them. Establishing protocols and practices,
managing logistics, establishing trust — all these matter. It is,
Healy is arguing, not useful to talk about the issue or organ transplants without
addressing the specifics of questions such as whether relatives of dead
organ donors have the right to meet the recipient (as justone example), because these are
the kind of decisions that can have big consequences.
The book is
academically written, with all the costs and benefits that implies. It
was originally a PhD thesis, so unsurprisingly it favours logic and
cautious language over passion, footnotes everything, and tends to wrap
conclusions in qualifiers. The empirical middle chapters in particular
(3 and 4) are a bit dry. But the book is an important contribution; it identifies a whole range of
important issues when it comes to organ donation, and may move the
debate away from the market/altruism dichotomy to a more nuanced and realistic debate
over coping with large-scale enterprises. That would be valuable. I definitely recommend it if you
want to learn about the issues involved, and want some ideas for
healthy ways forward.
(A minor quibble before moving on. He discusses the US and a bunch of European countries, but not Canada. This is common in American/European comparisons — The Economist seems prone to it — but it is still irritating to this Anglo-Canadian reader.)
Healy looks at both the blood system and
the organ system, and his analysis of the failure of the blood system
in the US and in Europe to handle HIV and Hepatitis C infection makes
his reality-based approach uncomfortable reading for market enthusiasts
and market sceptics alike. In some cases, the market-based blood plasma
system in the US did better than the donation-based mainstream blood
system in responding to the potential for infection, even as both sets
of organizations had the same information at the same time. That they
too failed at the hurdle of throwing out blood plasma that was already
taken is no comfort. In fact, one of the more interesting conclusions
that Healy seems to come to is that there may be dependencies among the
different parts of the system (the collection agency, the donors, the
recipients, the hospitals) that are more important than the presence of
absence of payment in the determination of success or failure. The
issue may be more one of industrialization of the system than the
commodification of organs. This makes sense, once he points it out,
because any market-based approach is going to involve local monopsonies
and a few big players in any given region. There’s not going to be a
whole lot of competition going on among agencies, offering higher and
lower prices for kidneys, and so the outcome will depend on the
detailed interactions that take place. And if the shortage is to be
tackled (there were 50,000 people on the waiting list for kidneys in
the USA in 2000) then this is going to be a large-scale, industrial
effort whether or not payment is made to "donors" or not. Money will
be involved because big organizations – private industry or not – have
lots of money flowing through them. People will make or break their
careers based on what decisions are to be made, and as the
HIV/hepatitis case proved, a public or not-for-profit agency is no
proof against tragedy.
Much of the discussion is handled in
terms of literature on the "gift relationship". It’s not something I
know much about (the literature, not the relationship, although I am a
bit cheap), but it seems to handle what sociology is best at. There are
layers of meaning that influence our decisions and attitudes to issues
such as organ transplants. Healy reminds us that life insurance was
once a controversial industry, with its connotations of payment for
death, that had to overcome cultural resistance and find ways to stake
a position that is both morally acceptable and profitable. The life
insurance industry was, of course, one of the last havens of
large-scale mutual co-operative organizations, and in some countries
the movement away from that model to a shareholder model is one that
has not yet played out. But I digress.
The nature of the gift
relationship is subtle. If payment is made to a funeral home rather
than directly to family members, then does that mean the organs of a
just-deceased loved one have been donated (with an acknowledgement made
in honour of the gesture) or have the family been paid? If a live
kidney donor is compensated for their time and discomfort, but not for
their kidney, have they been paid? There are more subtleties in the
book, both involving payment and not. It seems that we all agree that
organ donation is good in the abstract, but not so much when it comes
to the crunch. Attempts to narrow this gap between abstract approval
and on-the-spot reluctance may be seen as delicate and considerate
diplomacy, or may be seen as cynical manipulation (Healy compares parts
of the process to the con-man’s efforts to "cool the mark" meaning to
make the object of a scam accept his/her position as loser in a
resigned manner rather than in anger, so that they don’t report it to
the authorities).
My guess is that the systems developed in
different countries will involve some forms of payment but will steer
clear of the obvious payment that Becker appears to advocate. Any
agency, public or private, will have to be monitored (oops – there’s
that passive voice), and that monitoring will cost money. Issues like
deciding on how to determine death (and debates are going on about this
now, of course) are vulnerable to all kinds of incentives I’d rather
not think about; but someone has to, and it shouldn’t be someone with a
direct monetary stake in the outcome.
Healy convinced me that the big
issue is not the economists’ issue — of markets versus altruism —
but is the sociologists’ issue of coping with complex incentives in
large-scale industrial organizations, and that alone was worth the price of the book. Recommended.
john_d at This Magazine writes about Canadian wines, and what makes them (or fails to make them) Canadian. The story is similar to that of British beer.
Link: BLOG.THISMAGAZINE.CA.
Went into Toronto today, and took a look at the Andy Warhol exhibition at the Art Gallery of Ontario, "guest curated" by David Cronenberg.
Talking to AK beforehand, she said that the show is as much about
Cronenberg as it is about Warhol, and I see what she means. The theme
is "Stars, Deaths and Disasters, 1962–1964", and it’s
the Deaths part that is most obvious, and which Cronenberg has most to
say about on the commentary. Given that Warhol produced so much stuff
the focus on death, particularly grisly death (as in the electric chair
series and the Car Crash series) and disaster does seem to have more to
do with Cronenberg’s interests than with Warhol’s.
I went not knowing much about Warhol beyond Campbell Soup, Velvet
Underground, and the odd profile of him I’d seen. I wasn’t sure whether
I would be impressed or not – I don’t have knee jerk reactions for or
against modern art. On the plus side, I like being challenged by art –
anything that makes you take a different look at a piece of the world
deserves praise, whether you agree with it or not – and on the negative
side, I don’t have much time for sensationalism for the sake of it. But
how to tell the difference?
I’m glad I saw the exhibition, and some of the things I really liked
(the Elvis image, for example) but overall I ended up with a lower
opinion of Warhol than when I went in.
The films were the least impressive part of the exhibition. ("Blow
Job", "Sleep", "The Couch", "Screen Tests", and something about a
haircut) There is always something about yesterday’s iconoclasts that
is a little pathetic, because the most outrageous things tend to look
tamer over time (well, except for Un Chien Andalou perhaps). Most
people go through a phase of self-discovery and exploration of our
place in the world, some with more gusto than others. But most of us
don’t call it ground-breaking art and I didn’t see much in the films
beyond a desire to shock and a desire to self promote. The expressions
on some of the models/participants/actors were just "hey, look at us,
aren’t we something" and I thought – "no". The films seemed to catch
the worst of what the Warhol phenomenon is about: the circular
reasoning behind the fame and celebrity that he seemed to pursue so
relentlessly. Warhol is important in part because of the subjects of
his art (Jackie O, Liz Taylor, Marilyn Monroe); but in some cases the
subjects take on their importance only because it was Warhol who
pointed his camera at them: Warhol, in the end, is important only
because he is Warhol. The insights that the commentary gives into his
apparent shyness, his pursuit of celebrity, and his devotion to
celebrity are creepy. There is a touch of the Paris Hilton here –
famous for being famous. And if Warhol, you say, is deeper than Paris
Hilton, then he would disagree – one wall had his epigram on it: "If you want to know all about Andy Warhol, just look at the surface of
my paintings and films and me, and there I am. There’s nothing behind
it." The two of them share something in common.
The silk-screen images were more interesting, and occasionally more
disturbing. The best — the electric chair series (can’t remember their
real name) and some of the Liz Taylor series, and the famous Elvis
Presley images, use the repetitive silk-screen technique to fine
effect, with the sequence of images fading away, or collapsing on
themselves, giving a poignant and melancholy air to the whole image.
Others, such as Race Riot and Car Crash, disturbed me for different
reasons. Where the commentary argued that Warhol forces us to look in a
different way at the images, I’m afraid I just saw them as a
self-promotional artist taking others’ grief and distress and making
himself famous from it. His own distance from his subject did not have
the effect on me that it had on Cronenberg. He found the distancing
effect of silkscreen, the coldness of the technique, to demand a new
scrutiny of the image. I had no such reaction – to me Warhol’s
recycling of these images as art had little impact.
I’m glad I went, if only to see the iconic Presley image at full size
and in the silk-screened flesh – starting tall and bold, and fading
away into a dim greyness over time, it’s difficult not to see it as
prophecy. But I can’t take Warhol seriously as a major artist. The fact
that his reputation has grown since his death is, I suspect, mainly a
result of his contemporaries bringing sentimental memories of the their
youth into the now top-of-the-field positions that they occupy.
Scott Piatkowski highlights a paragraph from the local paper that reminds us how market-driven and developer-driven much of Southern Ontario’s recent city growth has been.
"Historically, in the early 90s, (the idea) was to strip away bureaucracy to make it easy for business to come down and open up without all the red tape," city planner Cory Bluhm said yesterday. "What we’ve learned from this is the opposite is now the way to go. Residents now expect the city to play a role in shaping the downtown."
About time, I’d say. The only thing worse than city hall is no city hall. Here is the full report that he is talking about.
