The changing economics of spam

An interesting Wall Street Journal story today on the “Buffalo Spammer”, a fellow who was rotating through multiple Earthlink accounts sending millions of spam messages. Apparently the WSJ story was very well-timed: he was hit with $14mm in fines late today.

That is, of course, interesting in and of itself. But more interesting, to me anyway, is the changing economics this will create for spammers.

Historically spam has been a free-rider on the email networks: the rest of us have paid the freight for email users who flooded network capacity far in excess of their monthly charges. There were many solutions, but ISPs were uninterested in them, things like charging per megabyte, charging for over a certain number of email messages being sent per day/week/month, and so on.

Instead ISPs responded by kicking users off their networks. In effect, they forced spammers to open new accounts with new stolen credit cards or forged identities (or both). This was sometimes effective in the short term, but it wasn’t much of a long-run deterrence. Specifically, it imposed negligible costs, and it didn’t eliminate the free-rider problem. It was sort of like kicking vandals out of your house rather than prosecuting them.

Now, however, with Earthlink successfully bringing suit against a profligate spammer, it will almost certainly help to make those exogenous costs endogenous for the spammer. They must worry that, if caught, they will face similarly high penalties, rather than simply being kicked out for sending emails.

But there is still one worry: look at how much effort was involved on Earthlink’s part in tracking down this one spammer. It was very, very costly. While it wasn’t “$14mm costly”, it was certainly at least a million dollars. Small ISPs will rarely do a Clifford Stoll and seek such interlopers out, so it is unclear to me, in the absence of better email tracking, how effective Earthlink’s action will end up being.

SARS update: Lancet data, etc.

A SARS update: As I have been saying, the outbreak is now largely a China problem, as the following graph shows (full data is here):


Most significant recent news was the (marginally flawed) Lancet study showing that mortality rates are much higher than early reports led many to think. Far from being percentages in the low single digits, they are actually more like 55 per cent in infected people over 60 years old, and 13 per cent in infected people aged under 60. The study also showed that a reasonable incubation period was more like 14 days than 10, as is currently used. Finally, study showed the remarkably long hospital stays associated with the disease, as the following figure shows:


Why is this interesting? Because it shows the societal and healthcare costs imposed by an outbreak. It would be easy to imagine how a concentrated outbreak could quickly fill all beds and impose serious economic consequences on a system ill-equipped to cope.

The trouble with CSS

Geek note: Great, savvy comments from Tim Bray on the pluses and minuses of dealing with Cascading Style Sheets (CSS). Like democracy, they’re better than the alternatives, but that’s not necessarily saying much.

Two key points that he makes:

  • CSS is harder to maintain than either content or computer programs. Personally, I find them nightmarish, and it’s so-so easy for a CSS document to descend into meaninglessness.
  • When you make a change and the resulting CSS doc blows up your page, there is really no straightforward way to debug the causal chain. Had that happen the other day for this site and finally had to regress back to a known working copy, despite making only minor changes.

Venture capital continues the slide

Venture capital is rapidly becoming a derelict industry. News this week that a partner in a major Boston fund recently moved out to play “house” in South Carolina. And more news today that the first quarter of the year followed recent trends: the amount of money has declined every quarter since early 2000.

Other facts:

  • There has been one public-offering for a venture-backed company since January of 2003, down from 262 in the same period during 2000
  • There were 623 venture deals in Q1, the smallest number since 1996
  • Workshops on obtaining goverment grants are now a growth business

SARS update

Eerie how the peak of media SARS frenzy matched the moment when the outbreak went into serious decline. The Time/Newsweek contrarian indicator strikes again!

More seriously, as I have been saying for some time now, the latest data shows that SARS is under control in all regions, with virtually no new cases outside of China. Even the exception, China, is having some success controlling the outbreak.

On an investment front, SARS shorts who didn’t cover some time back, as I suggested two weeks or so ago, are taking it in the teeth. And SARS longs who are playing with various micro-cap biotechs are hanging by their teeth instead.

Finally, as a general note on biotech, this morning’s developments were interesting. There was a classic tragedy of the anti-commons, as patents were filed by Hong Kong, Canada, and U.S. agencies on aspects of the virus and related genes. All would seemingly have preferred not to file, but all also worried that someone else would and they would be forced to pay a ticket-taker for access to the SARS market — assuming it becomes a persistent or seasonal syndrome.

Those smarty-pants at Mensa

Somewhat off-topic, but those smarty-pants at Mensa have always irritated me more than a little. It took, however, a comment on Peter Hall’s site to really get me thinking about just how kooky they are.

Peter points to this Mensa page and says that with all the high IQs in their supposed target audience, is it really necessary that Mensa tell would-be applicants that 1 in 50 is the same as 2%?

Fair comment, and I would add to that something else further up the page:

Many people are surprised to learn they have an above average IQ.
Yes, given that half the population is by definition above average, I would guess …. oh, about half the population is “surprised”.

Revving up the technology IPO engine

Much is being made of two tech IPOs coming out this week. Yes, the arrival of Diginet and iPayment (oh, that’s so 1990s sounding!) will end a two-month lull in the offering market. While the underwriters are saying that the road show is going well (what else would you expect them to say), Diginet and iPayment are both losing money.

What happens if both or either do well? Near-term, I count 18 U.S. IPOs in the backlog, and they’d happily come to market, looking to raise about $3.6 billion.

But the sumo wrestler in the closet is, of course, Google. The parade of stories about that company’s potential offering is growing, and the company is participating in some of the stories. And as I pointed out on the weekend, ex-analyst Lise Buyer has joined the company in a position that I would call help-tridy-us-up-to-go-public.

What would a Google offering look like? Well, with something like $500-mm in sales this year, and cashflow positive since December of 2000, it would be the first tech offering in some time to attract broad and deep institutional interest.

Is truth in accounting an oxymoron?

Tim Bray repeats the usual criticisms of GAAP (generally accepted accounting principles). While he’s right, of course — GAAP is overly subjective — that shouldn’t be news to anyone. After all, its inherent subjectivity is right there in the acronym if you choose to look. The principles are generally accepted, not universally accepted, not laws of nature.

In my experience, engineers and scientists always have a hard time with the idea that accounting statements allow management to have some discretion about when the recognize revenues and earnings. As Tim does here, they usually turn to cash flow statements, suggesting those are closer to the truth. While they are one type of truth — what is happening with cash — they too distort the investments a company makes and when it can expect to earn a return on those investments.

So, cash flow statements are fine, but they are not strictly better than accounting earnings statements. It is a fiction to pretend otherwise. Scan both, and don’t get hung up on some misguided search for truth in accounting.

Frontline discovers Wall Street (again)

May 8th will see a no-doubt downbeat report by Frontline on the brokerage industry, Worldcom, and Wall Street’s latest escapades. I’m not sure what more there is left to say on this subject, and snippets in Frontline’s press prelease promoting the program don’t give me oodles of hope.

Case in point, the following snippet:

“Is Wall Street fixed? Ha! Not by a long shot,” investor George Mickelis says. Mickelis, the owner of a family restaurant in Houston, lost $300,000 that he invested in telecom stocks for his children’s college education. “I have no faith and no confidence–not in the companies, not in the brokerage firms…. What have they done to restore my faith and confidence as an individual investor? Zero. Zip.”

While that may sound sad and piquant to some, it is entirely indefensible and unsympathetic. This fellow put a meaningful component of money into telecommunications’ stocks in saving for his childrens’ education? And then he expects sympathy?

That is an outrageous and entirely silly claim. Anyone who puts money into a) equities, and b) more specifically, technology equities saving for a known, near-term deadline needs psychiatric help, not legal redress for lost principal.

Venture data is lossy

After much tussling with venture funds and (mostly) the media, CalPERS finally released performance data for its venture capital investments. Here is that data in a more useful form:

Year Funds Average St dev
1993 5 19% 3.6%
1994 18 13% 6.0%
1995 14 17% 8.8%
1996 17 6% 5.2%
1997 13 11% 7.1%
1998 21 -3% 3.8%
1999 12 -8% 5.4%
2000 29 -15% 9.8%

Basically, you can see how returns deteriorate rapidly once you get past 1997. While it is a reasonable excuse to say that funds in their early years typically show poor returns, 1998 funds are now five years old and should be showing better than break-even numbers.

One other point: the standard deviation of returns is remarkably high. For vintage 2000 funds, the table shows that a 95% confidence interval (assuming return normality) would range from almost -35% to +5% — a very large range.

Here is the same data for CalPERS’ venture investments:




St dev