Given this, it appears that an open source venture (a company that can scale to millions of worker/owners creating a new economic ecosystem) that builds massive human curated databases and decentralizes the processing load of training these AIs could become extremely competitive.
But what if the economic ecosystem could exist
without the venture? Instead of trying to build a
virtual company with millions of workers/owners,
build a market economy with millions of participants
in tens of thousands of projects and tasks? All of
this stuff scales technically much better than it
scales organizationally—you could still be
part of a large organization or movement while only
participating directly on a small set of issues at
any one time. Instead of holding equity in a large
organization with all its political risk, you could
hold a portfolio of positions in areas where you have
enough knowledge to be comfortable.
Robb's opportunity is in training AIs, not in writing
code. The "oracle" for resolving AI-training or
dataset-building contracts would have to be different,
but the futures market could be the same.
The cheating project problem
Why would you invest in a futures contract on bug
outcomes when the project maintainer controls the
And what about employees who are incentivized from
both sides: paid to fix a bug but able to buy futures
contracts (anonymously) that will let them make more
on the market by leaving it open?
In order for the market to function, the total
reputation of the project and contributors
must be high enough that outside participants
believe that developers are more motivated
to maintain that reputation than to "take a
on a bug.
That implies that there is some kind of relationship
between the total "reputation capital" of a project
and the maximum market value of all the futures
contracts on it.
Open source metrics
To put that another way, there must be some
relationship between the market value of futures
contracts on a project and the maximum reputation
value of the project. So that could be a proxy for
a difficult-to-measure concept such as "open source
Open source journalism
Hey, tickers to put into stories! Sparklines! All
the charts and stuff that finance and sports reporters
can build stories around!
This paper presents the largest study to date on gender bias, where we compare acceptance rates of contributions from men versus women in an open source software community. Surprisingly, our results show that women's contributions tend to be accepted more often than men's. However, women's acceptance rates are higher only when they are not identifiable as women.
For outsiders, women coders who use gender-neutral profiles get their changes accepted 2.8% more of the time than men with gender-neutral profiles, but when their gender is obvious, they get their changes accepted 0.8% less of the time.
The experiment, launching this month,
will help reviewers who want to try breaking
habits of unconscious bias (whether by gender or
insider/outsider status) by concealing the name and
email adddress of a code author during a review on
Bugzilla. You'll be able to un-hide the information
before submitting a review, if you want, in order
to add a personal touch, such as welcoming a new
The extension will "cc" one of two special accounts
on a bug, to indicate if the review was done partly
or fully blind. This lets us measure its impact without
having to make back-end changes to Bugzilla.
(Yes, WebExtensions let you experiment
with changing a user's experience of
a site without changing production web
applications or content sites. Bonus link:
A first release is on a.m.o., here: Blind Reviews BMO
if you want an early look. We'll send out
notifications to relevant places when the "last"
bugs are fixed and it's ready for daily developer use.
Adfraud is a big problem, and we keep seeing two
basic approaches to it.
Flight to quality: Run ads only on trustworthy
sites. Brands are now playing the fraud game
with the "reputation coprocessors" of the
audience's brains on the brand's side. (Flight
to quality doesn't mean just advertise on the
same major media sites as everyone else—it
can scale downward with, for example, the Project
model that lets you choose sites that are "brand safe"
Increased surveillance: Try to fight adfraud by
continuing to play the game of trying to get big-money
impressions from the cheapest possible site, but throw
more tracking at the problem. Biggest example of this
is to move ad money to locked-down mobile platforms
and away from the web.
Anyway, I'm interested in and optimistic about
the results of the recent Mozilla/Caribou Digital
It turns out that USA-style adtech is harder to do
in countries where users are (1) less accurately
tracked and (2) equipped with blockers to avoid
bandwidth-sucking third-party ads. That's likely
to mean better prospects for ad-supported news and
cultural works, not worse. This report points
out the good news that the so-called adtech
tax is lower in developing countries—so
what kind of ad-supported businesses will be
enabled by lower "taxes" and "reinvention, not
of more magazine-like advertising?
Of course, working in those markets is going to be
hard for big US or European ad agencies that are
now used to solving problems by throwing creepy
tracking at them. But the low rate of adtech
taxation sounds like an opportunity for creative
local agencies and brands. Maybe the report should
have been called something like "The Global South
is Shitty-Adtech-Proof, so Brands Built Online There
Are Going to Come Eat Your Lunch."
Why would you want the added complexity of a
market where anyone can take either side of a
futures contract on the status of a software bug,
and not just offer to pay people to fix bugs like a
sensible person? IMHO it's worth trying not just
because of the promise of lower transaction costs
and more market liquidity (handwave) but because it
enables other kinds of transactions. A few more.
Partial work I want a feature, and buy the
"unfixed" side of a contract that I expect to lose.
A developer decides to fix it, does the work, and
posts a pull request that would close the bug.
But the maintainer is on vacation, leaving her
pull request hanging with a long comment thread.
Another developer is willing to take on the political
risk of merging the work, and buys out the original
Prediction/incentivization With the right market
design, a prediction that something won't happen
is the same as an incentive to make it happen.
If we make an attractive enough way for users
to hedge their exposure to lack of innovation,
we create a pool of wealth that can be captured
by innovators. (Related: dominant assurance
Bug triage Much valuable work on bugs is in
the form of modifying metadata: assigning a bug
to the correct subsystem, identifying dependency
relationships, cleaning up spam, and moving invalid
bugs into a support ticket tracker or forum.
This work is hard to reward, and infamously hard to
find volunteers for. An active futures market could
include both bots that trade bugs probabilistically
based on status and activity, and active bug triagers
who make small market gains from modifying metadata
in a way that makes them more likely to be resolved.
Content Neutrality: Content blocking software
should focus on addressing potential user needs
(such as on performance, security, and privacy)
instead of blocking specific types of content
(such as advertising).
Transparency & Control: The content blocking
software should provide users with transparency and
meaningful controls over the needs it is attempting
Openness: Blocking should maintain a
level playing field and should block under the
same principles regardless of source of the
content. Publishers and other content providers
should be given ways to participate in an open Web
ecosystem, instead of being placed in a permanent
penalty box that closes off the Web to their
products and services.
[T]he police are the public and that the public are the police, the police being only members of the public who are paid to give full-time attention to duties which are incumbent on every citizen in the interests of community welfare and existence.
Web browser developers have similar responsibilities
to those of Peel's ideal police: to build a browser
to carry out the user's intent, or, when setting
defaults, to understand widely held user norms and
implement those, while giving users the affordances
to change the defaults if they choose.
The question now is how to apply content blocking
principles to today's web environment. Some qualities
of today's situation are:
Tracking protection often doesn't have to be
perfect, because adfraud. The browser can provide
some protection, and influence the market in a
positive direction, just by getting legit users
below the noise floor of fraudbots.
Tracking protection has the potential to intensify
a fingerprinting arms race that's already going on,
by forcing more adtech to rely on fingerprinting
in place of third-party cookies.
Fraud is bad, but not all anti-fraud is good.
Anti-fraud technologies that track users can create
the same security risks as other tracking—and
enable adtech to keep promising real eyeballs on
crappy sites. The "flight to quality" approach
to anti-fraud does not share these problems.
Adtech and adfraud can peek at Mozilla's homework,
but Mozilla can't see theirs. Open source projects
must rely on unpredictable users, not unpredictable
platform decisions, to create uncertainty.
Which suggests a few tactics—low-risk ways to
apply content blocking principles to address today's
Empower WebExtensions developers and users. Much
of the tracking protection and anti-fingerprinting
magic in Firefox is hidden behind preferences. This
makes a lot of sense because it enables developers
to integrate their work into the browser in parallel
with user testing, and enables Tor Browser to do
less patching. IMHO this work is also important
to enable users to choose their own balance between
privacy/security and breaking legacy sites.
Inform and nudge users who express an interest
in privacy. Some users care about privacy, but
don't have enough information about how protection
choices match up with their expectations. If
a user cares enough to turn on Do Not Track,
change cookie settings, or install an ad blocker,
then try suggesting a tracking protection setting
or tool. Don't assume that just because a user
has installed an ad blocker with deceptive privacy
that the user would not choose privacy if asked
Understand and report on adfraud. Adfraud
is more than just fake impressions and clicks.
New techniques include attribution fraud: taking
advantage of tracking to connect a bogus ad impression
to a real sale. The complexity of attribution models
makes this hard to track down. (Criteo and Steelhouse
settled a lawsuit about this before discovery could
A multi-billion-dollar industry is devoted to
spreading a story that minimizes adfraud, while
independent research hints at a complex and lucrative
adfraud scene. Remember how there were two Methbot
Methbot got a bogus block of IP addresses, and Methbot
circumvented some widely used anti-fraud scripts.
The ad networks dealt with the first one pretty
quickly, but the second is still a work in progress.
The more that Internet freedom lovers can help
marketers understand adfraud, and related problems
such as brand-unsafe ad placements, the more that
the content blocking story can be about users, legit
sites, and brands dealing with problem tracking,
and not just privacy nerds against all web business.
More on the third connection in Benkler’s
which was pretty general. This is just some notes
on more concrete examples of how new kinds of direct
connections between markets and peer production might
work in the future.
Smart contracts should make it possible to enable
these in a trustworthy, mostly decentralized, way.
Feature request I want emoji support on my blog,
so I file, or find, a wishlist bug on the open source
blog package I use: "Add emoji support." I then offer
to enter into a smart contract that will be worthless
to me if the bug is fixed on September 1, or give me
my money back if the bug is unfixed at that date.
A developer realizes that fixing the bug would be
easy, and wants to do it, so takes the other side
of the contract. The developer's side will expire
worthless if the bug is unfixed, and pay out if the
bug is fixed.
"Unfixed" results will probably include bugs that
are open, wontfix, invalid, or closed as duplicate
of a bug that is still open.
"Fixed" results will include bugs closed as fixed,
or any bug closed as a duplicate of a bug that is
closed as fixed.
If the developer fixes the bug, and its status changes
to fixed, then I lose money on the smart contract but
get the feature I want. If the bug status is still
unfixed, then I get my money back.
So far this is just one user paying one developer to
write a feature. Not especially exciting. There is
some interesting market design work to be done here,
though. How can the developer signal serious interest
in working on the bug, and get enough upside to be
meaningful, without taking too much risk in the event
the fix is not accepted on time?
Arbitrage I post the same offer, but another user
realizes that the blog project can only support emoji
if the template package that it depends on supports
them. That user becomes an arbitrageur: takes the
"fixed" side of my offer, and the "unfixed" side of
the "Add emoji support" bug in the template project.
As an end user, I don't have to know the dependency
relationship, and the market gives the arbitrageur
an incentive to collect information about multiple
dependent bugs into the best place to fix them.
Front-running Dudley Do-Right's open source
project has a bug in it, users are offering to buy the
"unfixed" side of the contract in order to incentivize
a fix, and a trader realizes that Dudley would be
unlikely to let the bug go unfixed. The trader takes
the "fixed" side of the contract before Dudley wakes
up. The deal means that the market gets information
on the likelihood of the bug being fixed, but the
developer doing the work does not profit from it.
This is a "picking up nickels in front of a
steamroller" trading strategy. The front-runner is
accepting the risk of Dudley burning out, writing a
long Medium piece on how open source is full of FAIL,
and never fixing a bug again.
Front-running game theory could be interesting. If
developers get sufficiently annoyed by front-running,
they could delay fixing certain bugs until after the
end of the relevant contracts. A credible threat to
do this might make front-runners get out of their
positions at a loss.
CVE prediction A user of a static analysis tool
finds a suspicious pattern in a section of a codebase,
but cannot identify a specific vulnerability. The
user offers to take one side of a smart contract
that will pay off if a vulnerability matching a
certain pattern is found. A software maintainer or
key user can take the other side of these contracts,
to encourage researchers to disclose information and
focus attention on specific areas of the codebase.
Security information leakage Ernie and Bert
discover a software vulnerability. Bert sells it to
foreign spies. Ernie wants to get a piece of the
action, too, but doesn't want Bert to know, so he
trades on a relevant CVE prediction. Neither Bert nor
the foreign spies know who is making the prediction,
but the market movement gives white-hat researchers
a clue on where the vulnerability can be found.
Open source metrics: Prices and volumes on bug
futures could turn out to be a more credible signal
of interest in a project than raw activity numbers.
It may be worth using a bot to trade on a project you
depend on, just to watch the market move. Likewise,
new open source metrics could provide useful trading
strategies. If sentiment analysis shows that a
project is melting down, offer to take the "unfixed"
side of the project's long-running bugs? (Of course,
this is the same market action that incentivizes
fixes, so betting that a project will fail is the
same thing as paying them not to. My brain hurts.)
What's an "oracle"?
The "oracle" is the software component that moves
information from the bug tracker to the smart
contracts system. Every smart contract has to be tied
to a given oracle that both sides trust to resolve
For CVE prediction, the oracle is responsible
for pattern matching on new CVEs, and feeding the
info into the smart contract system. As with all
of these, CVE prediction contracts are tied to a
Bots might have several roles.
Move investments out of duplicate bugs. (Take a "fixed"
position in the original and an "unfixed" position in the
duplicate, or vice versa.)
Make small investments in bugs that appear valid
based on project history and interactions by
Track activity across projects and social sites
to identify qualified bug fixers who are unlikely
to fix a bug within the time frame of a contract,
and take "unfixed" positions on bugs relevant
For companies: when a bug is mentioned in an
internal customer support ticketing system, buy
"unfixed" on that bug. Map confidential customer
needs to possible fixers.
Since most software is sold with an “as is” license, meaning the company is not legally liable for any issues with it even on day one, it has not made much sense to spend the extra money and time required to make software more secure quickly.
The software business is still stuck on the kind of
licensing that might have made sense in the 8-bit
micro days, when "personal computer productivity"
was more aspirational than a real thing, and software
licenses were printed on the backs of floppy sleeves.
Today, software is part of products that do real
stuff, and it makes zero sense to ship a real product,
that people's safety or security depends on, with the
fine print "WE RESERVE THE RIGHT TO TOTALLY HALF-ASS
OUR JOBS" or in business-speak, "SELLER DISCLAIMS
THE IMPLIED WARRANTY OF MERCHANTABILITY."
But what about open source and
collaboration and science, and all that
stuff? Software can be both "product" and
Should there be a warranty on speech? If I dig up my
for re-running the make command when a
source file changes, and put it on the Internet,
should I be putting a warranty on it?
It seems that there are two kinds of software: some is
more product-like, and should have a grown-up warranty
on it like a real busines. And some software is more
speech-like, and should have ethical requirements like
a scientific paper, but not a product-like warranty.
What's the dividing line? Some ideas.
"productware is shipped as executables, freespeechware
is shipped as source code" Not going to work for
elevator_controller.php or a home router security
"productware is preinstalled, freespeechware is downloaded
separately" That doesn't make sense when even
implanted defibrillators can update over the net.
"productware is proprietary, freespeechware is open
source" Companies could put all the fragile stuff
in open source components, then use the DMCA and
CFAA to enable them to treat the whole compilation
Software companies are built to be good at getting
around rules. If a company can earn all its money in
faraway Dutch Sandwich Land and be conveniently too
broke to pay the IRS in the USA, then it's going to
be hard to make it grow up licensing-wise without
hurting other people first.
How about splitting out the legal advantages that the
government offers to software and extending some to
productware, others to freespeechware?
license may disclaim implied warranty
no anti-reverse-engineering clause in a freespeechware license is enforceable
freespeechware is not a "technological protection measure" under section 1201 of Title 17 of the United States Code (DMCA anticircumvention)
exploiting a flaw in freespeechware is never a violation of the Computer Fraud and Abuse Act
If the license allows it, a vendor may sell freespeechware, or a
derivative work of it, as productware. (This
could be as simple as following the You may
charge any price or no price for each copy that
you convey, and you may offer support or warranty
protection for a fee. term of the GPL.)
license may not disclaim implied warranty
licensor and licensee may agree to limit reverse engineering rights
DMCA and CFAA apply (reformed of course, but that's another story)
It seems to me that there needs to be some kind of
quid pro quo here. If a company that sells software
wants to use government-granted legal powers to
control its work, that has to be conditioned on not
using those powers just to protect irresponsible
Check it out—I'm "on Facebook"
again. Just fixed my gateway through
dlvr.it. If you're reading
this on Facebook, that's why.
Dlvr.it is a nifty service that will post to
social sites from an RSS feed. If you don't
run your own linklog feed, the good news is that
Pocket will generate RSS feeds from the articles you
so if you want to share links with people still on
Facebook, the combination of Pocket and dlvr.it
makes that easy to do without actually spending
human eyeball time there.
There's a story about Thomas Nelson,
leader of the Virginia Militia in the Revolutionary
During the siege and battle Nelson led the Virginia Militia whom he had personally organized and supplied with his own funds. Legend had it that Nelson ordered his artillery to direct their fire on his own house which was occupied by Cornwallis, offering five guineas to the first man who hit the house.
Would Facebook's owners do the same, now that we
know that foreign interests use Facebook to subvert
America? Probably not. The Nelson story is just an
unconfirmed patriotic anecdote, and we can't expect
that kind of thing from today's post-patriotic
investor class. Anyway, just seeing if I can
move Facebook's bots/eyeballs ratio up a little.
I'm thankful that the sewing
was invented a long time ago, not today. If the
sewing machine were invented today, most sewing
tutorials would be twice as long, because all the
thread would come in proprietary cartridges, and you
would usually have to hack the cartridge to get the
type of thread you need in a cartridge that works
with your machine.
Tracking protection is still hard. You have to
provide good protection from third-party tracking,
which users generally don't want, without breaking
legit third-party services such as content delivery
networks, single sign-on systems, and shopping carts.
Protection is a balance, similar to the problem of
filtering spam while delivering legit mail. Just as
spam filtering helps enable legit email marketing,
tracking protection tends to enable legit advertising
that supports journalism and cultural works.
In the long run, just as we have seen with spam
filters, it will be more important to make protection
hard to predict than to run the perfect protection
out of the box. A spam filter, or browser, that
always does the same thing will be analyzed and
worked around. A mail service that changes policies
to respond to current spam runs, or an unpredictable
ecosystem of tracking protection add-ons
that browser users can install in unpredictable
combinations, is likely to be harder.
But most users aren't in the habit of installing
add-ons, so browsers will probably have to give them
a nudge, like Microsoft Windows does when it nags
the user to pick an antivirus package (or did last
time I checked.) So the decentralized way to catch
up to Apple could end up being something like:
When new tracking protection methods show up
in the privacy literature, quietly build the needed
browser add-on APIs to make it possible for
new add-ons to implement them.
Do user research to
guide the content and timing of nudges. (Some
prefer to be tracked, and should be offered a
chance to silence the warnings by affirmatively
choosing a do-nothing protection option.)
Help users share information about the pros and
cons of different tools. If a tool saves lots of
bandwidth and battery life but breaks some site's
comment form, help the user make the right choice.
Sponsor innovation challenges to incentivize
development, testing, and promotion of diverse
tracking protection tools.
Any surveillance marketer can install and test a
copy of Safari, but working around an explosion of
tracking protection tools would be harder. How to
set priorities when they don't know which tools will
What about adfraud?
Tracking protection strategies have to take adfraud
into account. Marketers have two choices for how to
deal with adfraud:
flight to quality
Flight to quality is better in the long run. But
it's a problem from the point of view of adtech
intermediaries because it moves more ad money to
high-reputation sites, and the whole point of adtech
is to reach big-money eyeballs on cheap sites.
Adtech firms would rather see surveillance-heavy
responses to adfraud. One way to help shift marketing
budgets away from surveillance, and toward flight
to quality, is to make the returns on surveillance
investments less predictable.
This is possible to do without making value
judgments about certain kinds of sites. If you like
a site enough to let it see your personal info,
you should be able to do it, even if in my humble
opinion it's a crappy site. But you can have this
option without extending to all crappy sites the
confidence that they'll be able to live on leaked
data from unaware users.
I have to admit that some people hate me, but I have to tell you something about hate. If sending an electronic advertisement through email warrants hate, then my answer to those people is "Get a life. Don't hate somebody for sending an advertisement through email." There are people out there that also like us.
According to spammers, spam filtering was just Internet
nerds complaining about something that regular users
actually like. But the spam debate ended when big
online services, starting with MSN, started talking
about how they build for their real users instead of
for Wallace's hypothetical spam-loving users.
If you missed the email spam debate,
don't worry. Wallace's talking
points about spam filters constantly get recycled by
surveillance marketers talking about tracking
But now it's not email spam that users supposedly
crave. Today, the Interactive Advertising Bureau
tells us that users want ads that "follow them around"
from site to site.
Enough background. Just as the email spam debate
ended with MSN's campaign, the third-party
web tracking debate ended on June 5,
With Intelligent Tracking Prevention, WebKit strikes a balance between user privacy and websites’ need for on-device storage. That said, we are aware that this feature may create challenges for legitimate website storage, i.e. storage not intended for cross-site tracking.
Surveillance marketers come up with all kinds of
hypothetical reasons why users might prefer targeted
ads. But in the real world, Apple invests time
and effort to understand user experience. When Apple
communicates about a feature, it's because that
feature is likely to keep a user satisfied
enough to buy more Apple devices. We can't read their
confidential user research, but we can see what the
company learned from it based on how they communicate
(Imagine for a minute that Apple's user research
had found that real live users are more like the
Interactive Advertising Bureau's idea of a user.
We might see announcements more like "Safari
automatically shares your health and financial
information with brands you love!" Anybody got one
of those to share?)
Saving an out-of-touch ad industry
Advertising supports journalism and cultural
works that would not otherwise exist.
It's too important not to save. Bob Hoffman
[H]ow can we encourage an acceptable version of online advertising that will allow us to enjoy the things we like about the web without the insufferable annoyance of the current online ad model?
The browser has to be part of the answer. If the
browser does its job, as Safari is doing, it can
play a vital role in re-connecting users with legit
advertising—just as users have come to trust
legit email newsletters now that they have effective
Safari's Intelligent Tracking Prevention is not the
final answer any more than Paul Graham's "A plan
for spam" was
the final spam filter. Adtech will evade protection
tools just as spammers did, and protection will have
to keep getting better. But at least now we can
finally say debate over, game on.
Looks like the spawn of Privacy Badger and cookie
double-keying, designed to balance user protection
from surveillance marketing with minimal breakage of
sites that depend on third-party resources.
(Now all the webmasters will fix stuff to make it
work with Intelligent Tracking Prevention, which
makes it easier for other browsers and privacy tools
to justify their own features to protect users.
Of course, now the surveillance marketers will rely
more on passive fingerprinting, and Apple has an
advantage there because there are fewer different
Safari-capable devices. But browsers need to fix
Apple does massive amounts of user research and
it's fun to watch the results leak through when
they communicate about features. Looks like they
have found that users care about being "followed"
from site to site by ads, and that users are still
pretty good at applied behavioral economics. The side
effect of tracking protection, of course, is that
it takes high-reputation sites out of competition
with the bottom-feeders to reach their own audiences,
so Intelligent Tracking Prevention is great news for
Meanwhile, I don't get Google's weak "filter"
Looks like a transparently publisher-hostile move
(since it blocks some potentially big-money
ads without addressing the problem of site
commodification), unless I'm missing something.
Benkler builds on the work of Ronald Coase, whose
The Nature of the Firm explains how transaction
costs affect when companies can be more efficient
ways to organize work than markets. Benkler adds
a third organizational model, peer production.
Peer production, commonly seen in open source
projects, is good at matching creative people to
As peer production relies on opening up access to resources for a relatively unbounded set of agents, freeing them to define and pursue an unbounded set of projects that are the best outcome of combining a particular individual or set of individuals with a particular set of resources, this open set of agents is likely to be more productive than the same set could have been if divided into bounded sets in firms.
Firms, markets, and peer production all have their
advantages, and in the real world, most productive
activity is mixed.
Managers in firms manage some production directly
and trade in markets for other production. This
connection in the firms/markets/peer production
tripod is as old as firms.
The open source software business is the second
connection. Managers in firms both manage software
production directly and sponsor peer production
projects, or manage employees who participate
But what about the third possible connection between legs of the tripod?
Is it possible to make a direct connection between
peer production and markets, one that doesn't go
through firms? And why would you want to connect peer
production directly to markets in the first place?
because that's where the money is, but because markets
are a good tool for getting information out of people,
and projects need information. Stefan Kooths,
Markus Langenfurth, and Nadine Kalwey wrote, in
"Open-Source Software: An Economic Assessment"
Developers lack key information due to the absence of pricing in open-source software. They do not have information concerning customers’ willingness to pay (= actual preferences), based on which production decisions would be made in the market process. Because of the absence of this information, supply does not automatically develop in line with the needs of the users, which may manifest itself as oversupply (excessive supply) or undersupply (excessive demand). Furthermore, the functional deficits in the software market also work their way up to the upstream factor markets (in particular, the labor market for developers) and–depending on the financing model of the open-source software development–to the downstream or parallel complementary markets (e.g., service markets) as well.
Because the open-source model at its core deliberately rejects the use of the market as a coordination mechanism and prevents the formation of price information, the above market functions cannot be satisfied by the open-source model. This results in a systematic disadvantage in the provision of software in the open-source model as compared to the proprietary production process.
The workaround is to connect peer production
to markets by way of firms. But the more that
connections between markets and peer production
projects have to go through firms, the more chances
to lose information. That's not because firms
are necessarily dysfunctional (although most are,
in different ways). A firm might rationally choose
to pay for the implementation of a feature that they
predict will get 100 new users, paying $5000 each,
instead of a feature that adds $1000 of value for
1000 existing users, but whose absence won't stop
them from renewing.
Some ways to connect peer production to markets
are already working. Crowdfunding for software
are furthest along, both offering support for
developers who have already built a reputation.
A decentralized form of connection is
which Balaji S. Srinivasan describes as a tradeable
version of API keys. If I believe that your network
service will be useful to me in the future, I can
pre-buy access to it. If I think your service will
really catch on, I can buy a bunch of extra tokens
and sell them later, without needing to involve you.
(and if your service needs network effects, now I
have an incentive to promote it, so that there will
be a seller's market for the tokens I hold.)
by Alexander Tabarrok, build on the crowdfunding
model, with the extra twist that the person proposing
the project has to put up some seed money that
is divided among backers if the project fails to
secure funding. This is supposed to bring in extra
investment early on, before a project looks likely
to meet its goal.
What happens when the software industry is forced to grow up?
I'm starting to think that finishing the tripod,
with better links from markets to peer production,
is going to matter a lot more soon, because of the
software quality problem.
Today's software, both proprietary and open
source, is distributed under ¯\_(ツ)_/¯ terms.
"Disclaimer of implied warranty of merchantability" is
lawyer-speak for "we reserve the right to half-ass our
jobs lol." As Zeynep Tufekci wrote in the New York
"The World Is Getting Hacked. Why Don’t We Do More
to Stop It?" At some point the users are going to
get fed up, and we're going to have to. An industry
as large and wealthy as software, still sticking to
Homebrew Computer Club-era disclaimers, is like a
40-something-year-old startup bro doing crimes and
claiming that they're just boyish hijinks. This whole
disclaimer of implied warranty thing is making us
look stupid, people. (No, I'm not for warranties
on software that counts as a scientific or technical
communication, or on bona fide collaborative development,
but on a product product? Come on.)
Grown-up software liability policy is coming,
but we're not ready for it. Quality software
is not just a technically hard problem. Today,
we're set up to move fast,
break things, and ship dancing pigs—with incentives
more powerful than incentives to build secure
software. Yes, you get the occasional DARPA
or tool to facilitate incremental
but most software is incentivized through too many
layers of principal-agent problems. Everything is
If governments try to fix software liability before
the software scene can fix the incentives problem,
then we will end up with a stifled, slowed-down
software scene, a few incumbent software companies
living on regulatory capture, and probably not much
real security benefit for users. But what if users
(directly or through their insurance companies) are
willing to pay to avoid the costs of broken software,
in markets, and open source developers are willing
to participate in peer production to make quality
software, but software firms are not set up to
What if there is another way to connect the "I would
rather pay a little more and not get h@x0r3d!" demand
to the "I would code that right and release it in open
source, if someone would pay for it" supply?
In the matter of reforming things, as distinct from deforming them, there is one plain and simple principle; a principle which will probably be called a paradox. There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, “I don’t see the use of this; let us clear it away.” To which the more intelligent type of reformer will do well to answer: “If you don’t see the use of it, I certainly won’t let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.
Bob Hoffman makes a good case for getting rid of user
in web advertising. But in order to take the next
steps, and not just talk among ourselves about things
that would be really great in the future, we first
need to think about the needs that tracking seems to
satisfy for legit marketers.
What I'm not going to do is pull out the
argument that's in every first comment on
every blog post that criticizes tracking: that
just technology and is somehow value-neutral.
Tracking, like all technologies, enables some kinds
of activity better than others. When tracking offers
marketers the opportunity to reach users based on
who the user is rather than on what they're reading,
watching, or listening to, then that means:
But if tracking is so bad, then why, when you go to
any message board or Q&A site that discusses marketing
for small businesses, is everyone discussing those
nasty, potentially civilization-extinguishing targeted
ads? Why is nobody popping up with a question on how
to make the next They Laughed When I Sat Down At the
Targeted ads are self-serve and easy to
get started with. If you have never bought
a Twitter or Facebook ad, get out your credit
card and start a stopwatch. These ads might be
but they have the lowest time investment of any legit
marketing project, so probably the only marketing
project that time-crunched startups can do.
Targeted ads keep your OODA loop tight. Yes,
running targeted ads can be addictive—If
you thought the the attention slot machine
on social sites was bad, try the advertiser
dashboard. But you're able to use them to
learn information that can help with the rest
of marketing. If you have the budget to exhibit
at one conference, compare Twitter ads targeted
to attendees of conference A with ads targeted to
attendees of conference B, and you're closer to
Marketing has two jobs: sell stuff to customers
and sell Marketing to management. Targeting is
great for the second one, since it comes with the
numbers that will help you take credit for results.
We're not going to be able to get rid of risky
tracking until we can understand the needs that it
fills, not just for big advertisers who can afford the
time and money to show up in Cannes every year, but
for the company founder who still has $1.99 business
and is doing all of Marketing themselves.
(The party line among web privacy people
can't just be that GDPR is going to
save us because the French powers that be
are all emmerdés ever since the surveillance/shitlord
tried to run a US-style game on their political
system. That might sound nice, but put not your trust
in princes, man. Even the most arrogant Eurocrats in
the world will not be able to regulate indefinitely
against all the legit business people in their
countries complaining that they can't do something
they see as essential. GDPR will be temporary air
for building an alternative, not a fix in itself.)
Post-creepy web advertising is still missing some key features.
Quick, low-risk service. With the exception
of the Project Wonderful
targeted ads are quick and low-risk,
while signal-carrying ads are the opposite.
A high-overhead direct ad sales process is not a
drop-in replacement for an easy web form.
I don't think that's all of them. But I don't
think that the move to post-creepy web advertising
is going to be a rush, all at once, either.
Brands that have fly-by-night low-reputation
competitors, brands that already have many
tracking-protected customers, and brands with
solid email lists are going to be able to move
faster than marketers who are still making tracking
work. More: Work together to fix web ads? Let's
I'm still two steps behind in devops
coolness for my network stuff. I don't even
have proper configuration management, and
that's fine because Configuration Management is an
now. Anyway, I still log in and actually run shell
commands on the server, and the LWN review of
mosh was helpful
to me. Now using mosh for connections that persist
across suspending the laptop and moving it from
network to network. More info: Mosh: the mobile
write a long Medium post apologizing to your users for failing
end date for IP Maximalism
When did serious "Intellectual Property Maximalism"
end? I'm going to put it at September 18,
which is the date that the Gates Foundation announced
funding for the Public Library of Science's
journal PLoS Neglected Tropical Diseases.
When it's a serious matter of people's health,
open access matters, even to the author of "Open
Letter to Hobbyists". Since then, IP Maximalism
stories have been mostly about rent-seeking
behavior, which had been a big part of the
freedom lovers's point all along. (Nobody quoted in this
story is pearl-clutching about "innovation",
for example: Supreme Court ruling threatens to
shut down cottage industry for small East Texas
Is it just me, or does it look to anyone else like the
man in the photo is checking the list of third-party
web trackers on the site to see who he can send a
National Security Letter to?
Could a US president who is untrustworthy
enough to be removed from office possibly be trustworthy
enough to comply with his side of a "Privacy
If it's necessary for the rest of the world to
free itself of its dependence on the U.S.,
does that apply to US-based Internet companies that
have become a bottleneck for news site ad revenue,
and how is that going to work?
If you're "verified" on
Twitter, you probably miss these, so I'll just
use my Fair Use rights to share that one with you.
Twitter is a uniquely influential medium, one that
shows up on the TV news every night and on news
sites all day. But somehow, the plan to make money
from Twitter is to run the same kind of targeted ads
that anyone with a WordPress site can. And the latest
Twitter news is a privacy update that includes, among
other things, more tracking of users from one site to
Yes, the same kind of thing that Facebook already
does, and better, with more users. And the same kind
of thing that any web site can already get from an
of companies. Boring.
If you want to stick this kind of ad on your
WordPress site, you just have to cut and paste some
ad network HTML—not build out a deluxe office
space on Market Street in San Francisco the way
Twitter has. But the result is about the same.
What makes Twitter even more facepalm-worthy is that
they make a point of not showing the ads to the
influential people who draw attention to Twitter to
start with. It's like they're posting a big sign
that says STUPID AD ZONE: UNIMPORTANT PEOPLE ONLY.
Twitter is building something unique, but they're
selling generic impressions that advertisers can get anywhere.
So as far as I can tell, the Twitter business model
is something like:
Money out: build something unique and expensive.
Money in: sell the most generic and shitty
thing in the world.
Facebook can make this work
because they have insane numbers of
Chump change per minute on Facebook still adds
up to real money. But Facebook is an outlier
on raw eyeball-minutes, and there aren't enough
minutes in the day for another. So Twitter
is on track to get sold for $500,000, like Digg
Which is good news for me because I know enough
Twitter users that I can get that kind of money
So why should you help me buy Twitter when you
could just get the $500,000 yourself? Because I have
a secret plan, of course. Twitter is the site that
everyone is talking about, right? So run the ads
that people will talk about. Here's the plan.
Sell one ad per day. And everybody sees the same one.
Sort of like the back cover of the magazine that
everybody in the world reads (but there is no such
magazine, so that's why this is an opportunity.)
No more need to excuse the verified users from
the ads. Yes, an advertiser will have to provide a
variety of sizes and localizations for each ad (and
yes, Twitter will have to check that the translations
match). But it's the same essential ad, shown to
every Twitter user in the world for 24 hours.
No point trying to out-Facebook Facebook
or out-Lumascape the Lumascape.
Targeted ads are weak on
and a bunch of other companies are doing them more
cost-effectively and at higher volume, anyway.
Of course, this is not for everybody. It's for
brands that want to use a memorable, creative ad to
try for the same kind of global signal boost that
a good Tweet® can get. But if you want generic
targeted ads you can get those everywhere else on the
Internet. Where else can you get signal? In order
to beat current Twitter revenue, the One Twitter
Ad needs to go for about the same price as a Super
Bowl commercial. But if Twitter stays influential,
that's reasonable, and I make back the 500 grand and a lot more.
Internet users have been asking what they can do to protect their own data from this creepy, non-consensual tracking by Internet providers—for example, directing their Internet traffic through a VPN or Tor. One idea to combat this that’s recently gotten a lot of traction among privacy-conscious users is data pollution tools: software that fills your browsing history with visits to random websites in order to add “noise” to the browsing data that your Internet provider is collecting.
[T]here are currently too many limitations and too many unknowns to be able to confirm that data pollution is an effective strategy at protecting one’s privacy. We’d love to eventually be proven wrong, but for now, we simply cannot recommend these tools as an effective method for protecting your privacy.
This is one of those "two problems one solution"
The problem for makers and users of "data
pollution" or spoofing tools is QA. How do you
know that your tool is working? Or are surveillance
marketers just filtering out the impressions
created by the tool, on the server side?
The problem for companies using so-called Non-Human
Traffic (NHT) is that when users discover
NHT software (bots), the users tend to remove
it. What would make users choose to participate
in NHT schemes so that the NHT software can run for
longer and build up more valuable profiles?
So what if the makers of spoofing tools could get a
live QA metric, and NHT software maintainers could
give users an incentive to install and use their
NHT market as a tool for discovering information
Imagine a spoofing tool that offers an easy way
to buy bot pageviews, I mean buy Perfectly
Legitimate Data on how fast a site loads from various
home Internet connections. When the tool connects
to its server for an update, it gets a list of URLs
to visit—a mix of random sites, popular sites,
and paying customers.
Now the spoofing tool maintainer will be able to to
tell right away if the tool is really generating
realistic traffic, by looking at the market price
of pageviews. The maintainer will even be able to
tell whose tracking the tool can beat, by looking
at which third-party resources are included on the
pages getting paid-for traffic.
The money probably won't be significant, since real
web ad money is moving to whitelisted, legit sites
and away from fraud-susceptible schemes anyway, but
in the meantime it's a way to measure effectiveness.
Setting up a couple of Linux systems to work
which is one of the things that I'm up to at work
FilterBubbler is a
and the setup instructions use
so I need NPM. In order to keep all the NPM stuff
under my own home directory, but still put the
web-ext tool on my $PATH, I need to make one-line
edits to three files.
One line in ~/.npmrc
prefix = ~/.npm
One line in ~/.gitignore
One line in ~/.bashrc
(My /bashrc has a bunch of export PATH= lines
so that when I add or remove one it's more likely to
get a clean merge. Because home directory in git.) I
think that's it. Now I can do
npm install --global web-ext
with no sudo or mess. And when I clone my home directory on another system it will just work.
(This is an answer to a question on
Twitter is the new blog comments (for now) and I'm
more likely to see comments there than to have time
to set up and moderate comments here.)
Adfraud is an easy way to make mad cash, adtech is
happily supporting it, and it all works because the
system has enough layers between CMO and fraud hacker
that everybody can stay as clean as they need to.
Users bear the privacy risks of adfraud, legit publishers pay for
and adtech makes more money from adfraud than fraud
hackers do. Adtech doesn't have to communicate or
coordinate with adfraud, just set up a fraud-friendly
system and let the actual fraud hackers go to work.
Bad for users, people who make legit sites, and civilization in
But one piece of good news is that adfraud can change
quickly. Adfraud hackers don't have time to get stuck
in conventional ways of doing things, because adfraud
is so lucrative that the high-skill players don't have
to stay in it for very long. The adfraud hackers
who were most active last fall have retired to run their
resorts or recording studios or wineries or whatever.
So how can privacy tools get a piece of the action?
One random idea is for an obfuscation tool
to participate in the market for so-called sourced
Fraud hackers need real-looking traffic and are
willing to pay for it. Supplying that traffic is
sketchy but legal. Which is perfect, because put
one more layer on top of it and it's not even sketchy.
And who needs to know if they're doing a good job
at generating real-looking traffic? Obfuscation tool
maintainers. Even if you write a great obfuscation
tool, you never really know if your tricks for helping
users beat surveillance are actually working, or if
your tool's traffic is getting quietly identified on
the server side.
In proposed new privacy tool model, outsourced QA pays YOU!
Set up a market where a Perfectly Legitimate Site
that is looking for sourced traffic can go to buy
pageviews, I mean buy Perfectly Legitimate Data on
how fast a site loads from various home Internet
connections. When the obfuscation tool connects to
its server for an update, it gets a list of URLs
to visit—a mix of random, popular sites and
Set a minimum price for pageviews that's high enough
to make it cost-ineffective for DDoS. Don't allow it
to be used on random sites, only those that the buyer
controls. Make them put a secret in an unlinked-to
URL or something. And if an obfuscation tool isn't
well enough sandboxed to visit a site that's doing
traffic sourcing, it isn't well enough sandboxed to
surf the web unsupervised at all.
Now the obfuscation tool maintainer will be able to
to tell right away if the tool is really generating
realistic traffic, by looking at the market price.
The maintainer will even be able to tell whose
tracking the tool can beat, by looking at which
third-party resources are included on the pages
getting paid-for traffic. And the whole thing can be
done by stringing together stuff that IAB members are
already doing, so they would look foolish to complain
If you want people on the Internet to argue with you, say that you're making a statement about values.
If you want people to negotiate with you, say that you're making a statement about business.
If you want people to accept that something is inevitable, say that you're making a statement about technology.
The mixup between values arguments, business
arguments, and technology arguments might be
why people are confused about Brands need to fire
by Doc Searls.
The set of trends that people call adtech is a
values-driven business transformation that is trying
to label itself as a technological transformation.
Some of the implementation involves technological
changes (NoSQL databases! Nifty!) but fundamentally
adtech is about changing how media business is
done. Adtech does have a set of values, none
of which are really commonly held even among people in
the marketing or advertising field, but let's not make
the mistake of turning this into either an argument
about values (that never accomplishes anything)
or a set of statements about technology (that puts
those with an inside POV on current technology at an
unnecessary advantage). Instead, let's look at the
business positions that adtech is taking.
Adtech stands for profitable
platforms, with commodity producers
of news and cultural works. Michael
Tiffany, CEO of advertising security firm White Ops,
saidThe fundamental value proposition of these ad tech
companies who are de-anonymizing the Internet is,
Why spend big CPMs on branded sites when I can
get them on no-name sites? This is not a
healthy situation, but it's a chosen path, not a
technologically inevitable one.
Adtech stands for making advertisers
support criminal and politically heinous
activity.I'll just let Bob Hoffman explain that
Fraudulent and brand-unsafe content is just the
overspray of the high value platforms/commoditized
content system, and advertisers have to accept
it in order to power that system. Or do they?
People have a lot of interesting decisions to make:
policy, contractual, infrastructural, and client-side.
When we treat the adtech movement as simply
technology, we take the risk of missing great
opportunities to negotiate for the benefit of brands,
publishers, and the audience.
This is a brand new blog, so I'm setting up
the basics. I just realized that I got the
whole thing working without a single script,
image, or HTML table. (These kids today
have it easy, with their media queries and CSS
One big question that I'm wondering about is: how many of the people
who visit here are using some kind of protection
from third-party tracking? Third-party tracking
has been an unfixed vulnerability in web browsers
for a long time. Check out the Unofficial Cookie
FAQ from 1997.
Third-party cookies are in there...and we're
still dealing with the third-party tracking problem?
In order to see how bad the problem is on this site,
I'm going to set up a little bit of first-party
data collection to measure people's vulnerability to
third-party data collection.
The three parts of that big question are:
Can a third-party tracker see state from other sites?
All it does is swap out the tracking image source three times.
When the Aloodo tracking script runs, to check if this browser is blocking the script from loading.
When the Aloodo script confirms that tracking is possible.
The work is done in the setupAloodo function,
which runs after the page loads. First, it sets the
src for the tracking pixel to js.png, then sets
up two callbacks: one to run after the Aloodo script
is loaded, and switch the image to ld.png, and
one to run if the script can track the user,
and switch the image to td.png.
Step three: check the logs
Now I can use the regular server logs to compare
the number of clients that load the original image,
load the two tracking images.
Metalsmith is pretty fun. The basic pipeline from
the article seems to work pretty well, but I ran
into a couple of issues. I might have solved these
in ways that are completely wrong, but here's what
works for me.
First, I needed to figure out how to get text from
an earlier stage of the pipeline. My Metalsmith
build is pretty basic:
turn Markdown into HTML (plus article metadata
apply a template to turn the HTML version into
a complete page.
That's great, but the problem seems to be with getting
a copy of just the HTML from step 1 for building the
index page and the RSS feed. I don't
want the entire HTML page from step 2, just the inner
HTML from step 1.
The solution seems to be
This doesn't actually strip off the template, just
lets you capture an extra copy of the HTML before
templatization. This goes into the pipeline after "markdown"
but before the "layouts" step.
Select Browse, then Browse Local, then select the .qcow2 file.
That's it. Now looking at a virtual MS-Windows guest
that I can use for those troublesome web conferences
(and for testing web sites under MSIE. If you try
the tracking test,
it should take you to a protection page that prompts
you to turn on the EasyPrivacy Tracking Protection
List. That's a quick and easy way to speed up your
web browsing experience on MSIE.)
Andrew Cowie has written something
The main thing that this one does differently is to
ask make which files matter to it, instead of doing
an inotifywatch on the whole directory. Comments and
The process is going to be a little different from
what you might be used to with another OS. If you
shop carefully (and reading blogs is a good first
step) then the drivers you will need are already
available through your Linux distribution's printer
HP has done a good job with enabling this.
The company has already released the necessary
printer software as open source, and your Linux
distribution has already installed it. So, go to
printers fully supported with the HPLIP software, pick a
printer you like, and you're done.
If you want a recommendation from me, the
HP LaserJet 3055,
a black and white all-in-one device,
has worked fine for me with various Linux setups
for years. It's also a scanner/copier/fax machine,
and you get the extra functionality for not much more
than the price of a regular printer. It also comes
with a good-sized toner cartridge, so your cost per
page is probably going to be pretty reasonable.
Other printer brands have given me more grief, but
fortunately the HP LaserJets are widely available
and don't jam much.
It's important not to show a smug expression on your
face while printing if users of non-Linux OSs are
still dealing with driver CDs or vendor downloads.
When you give travel directions, you include
landmarks, and "gone too far" points. Turn left after
you cross the bridge. Then look for my street and
make a right. If you go past the water tower you've
gone too far.
System administration instructions are much easier
to follow if they include those kind of check-ins
there, too. For example, if you explain how to set
up server software you can put in quick "landmark"
tests, such as, "at this point, you can run nmap and
see the port in the results." You can also include
"gone too far" information by pointing out problems
you can troubleshoot on the way.
A full-scale troubleshooting guide is a good idea,
but quick warning signs as you go along are helpful.
Much better than finding yourself lost at the end of
a long set of setup instructions.
doesn't accept dotted quads for ranges, but
fortunately most of the commands that accept an IP
address will also take it in the form of a regular
decimal. (Spammers used to use this to hide their
naughty domains from scanners that only looked for
the dotted quad while the browser would happily go to
So here's an ugly-ass shell function to convert an
IP address to a decimal. If you have a better one,
please let me know and I'll update this page. (Yes,
I know this would be one line in Perl.)
if [ $(echo $1 | grep -q '\.') ]; then
dq2int $(echo $1 | tr '.' ' ')
elif [ $# -eq 1 ]; then
total=$1; next=$2; shift 2
dq2int $(($total*2**8+$next)) $@
It says "Personal and Confidential" or "IMPORTANT
CORRESPONDENCE REGARDING YOUR OVERPAYMENT" on the
envelope—can you really discard it without
opening it? You sure can. Some junk mailers disguise
their mail pieces as important correspondence from
companies you actually do business with, and the
USPS helped them out a lot by renaming "Bulk Mail"
to "Standard Mail". But you can look at the postage
to discard "stealth" junk mail without opening it.
that any bills or mail containing specific
information about your business relationship with
the company must be mailed First Class.
So, if "Standard Mail" or "STD" appears in the upper
right corner, it's not a bill, it's not your new
credit card, and it's not a check. It's just sneaky
If you make a new ssh key and try to use it with ssh -i while running ssh-agent, ssh tries
the agent first. You could end up using a key provided by the agent
instead of the one you specify. You can fix this without killing
the agent. Use:
All that is really needed on computers
is a "Calculate" button or omnipresent menu command
that allows you to take an arithmetic expression,
like 248.93 / 375, select it, and do the calculation
whether in the word processor, communications package,
drawing or presentation application or just at the
Fortunately, there's a blue "Access
IBM" button on this keyboard that doesn't do much.
So, I configured tpb
"Access IBM" do this:
If you want to do this, besides tpb, you'll need xsel and xte, which is part of xautomation. If you don't have an unused button, you could also set up a binding in your window manager or build a big red outboard USB "eval" button or something.
The most important part of picking a distribution
is thinking about where you will go for help, and
what distribution that source of help understands.
That's true if your source of help is a vendor,
a consultant, or a users group.
If you're getting into uses for Linux that are
different from those of your local user group,
it's more important to use a list of people like
you than just the geographically closest user
group. For example, if you're planning to set
up a Linux-based recording studio and your local
LUG is all about running web sites and playing Crimson Fields, you might want to get on the
Planet CCRMA mailing list, and get your Linux
distribution recommendations there.
If you have a script that uses ssh, here's something
to put at the beginning of the script to make sure
the necessary passphrase has already been entered, and the
remote host is reachable, before starting a time-consuming
operation such as an rsync.