John Blair

this seemed like a 'economic theory' wrapped in a social context blanket.  Of course people are likely to go for more money - or at least the perception of it, whether or not they signal or screen behaviors.  Though the examples provided didn't seem overly relevant to the context our class, there are some ideas to incorporate into designs.

Signaling within a membership for instance might indicate that some of the members can't be trusted with private information, since they "leak" it to others outside the group for instance.  People always want to know about others secrets, so the audience is there for those wanting to disclose something they were asked not to reveal.

Screening would then allow the newly informed party to begin to dig deeper into the mystery by making inquiries of other members, by portraying him/her self as someone in the know, thus gaining further information from unsuspecting / naive members.

I did agree with the menu option mentioned in the reading where options are presented to a group and behavior is rewarded based on a self chosen selection.  This would take the sting out of the employer / community leaders delivering bad news as it should be obvious why the results were not rewarded better.

Signaling and screening is basically the simplified version of the principal/agent game. There isn't much to disagree about in this article. The author clearly explains that these theories are only theories, and that when considered as-is, misleading perspectives such as "Education is merely a signal and the community as a whole might benefit if it wasn't practiced" inevitably surface. Having said this, signaling is still incredibly important and often accurate.

Milgrom and Roberts identifies two general problems behind signaling, which basically discuss the issues:

1) People can lie about their "type", what stops them from going out of their comfort zone to obtain or display a signal.

2. By relying on certain signals and standards, false negatives are inevitable, meaning that just because some employees don't hold a Masters degree, it doesn't necessarily mean that they really are less productive then employees that do.

In general, these are good questions to consider, but as the authors pointed out, there are still broad assumptions. For instance, perhaps in the process of obtaining higher education, unproductive employes develop the skills and abilities that a highly productive employee might have. However, the reason that signaling is still pretty much the "rule of thumb" in many situations (academic, industry, or market-related), is that perhaps the current system is sustainable. Bigger and more profitable firms are still able to contract the highest-profiled celebrities, and individuals who strive for higher education are still in general the individuals that have the capabilities to perform well, or at least, commitment to a given cause. Of course, often times we observe mismatches, where a small firm uses all its resources to obtain a commercial slot during the Super Bowl, but failing to get the expected returns. Basically, my perspective is that "yes, lemons in a used-car market do indeed drive out the plums & peaches, but only in the long run". Perhaps the current signal system that we observe today is simply still at a early enough stage, which might adress the first problem proposed by Milgrom and Roberts.

Requirements on information processing are prohibitive and this is also knows as the problem of bounded rationality. Being the principal, it is usually hard to monitor the work of the agent who uses your expensive resources and somewhat determines your success. To solve this problem information is the key. The hidden action problem mainly comes from asymmetric information, thus if it is ever possible to design agreements that would motivate agents to self-monitor. And this is the key idea behind the economic example given in the reading. In Wikipedia, it gives the following definition:

principal-agent problem treats the difficulties that arise under conditions of incomplete and asymmetric information when a principal hires an agent. Various mechanisms may be used to try to align the interests of the agent with those of the principal, such as piece rates/commissions, profit sharing, efficiency wages, the agent posting a bond, or fear of firing. The principal-agent problem is found in most employer/employee relationships, for example, when stockholders hire top executives of corporations. Numerous studies in political science have noted the problems inherent in the delegation of legislative authority to bureaucratic agencies. Especially since bureaucrats often have expertise that legislators and executives lack, laws and executive directives are open to bureaucratic interpretation, creating opportunities and incentives for the bureaucrat-as-agent to deviate from the preferences of the constitutional branches of government. Variance in the intensity of legislative oversight also serves to increase principal-agent problems in implementing legislative preferences.

 To briefly summarize the economic model in MR’s reading, it tells:1.  Informativeness: structure pay to maximize informativeness

2.       Incentive Intensity: Optimal intensity depends on four factors:

--     incremental profits from additional effort

--     precision of estimating effort

--     agent’s risk tolerance

--     agent’s responsiveness to incentives

3.       Monitoring intensity: “ Spend more on improving monitoring when incentive intensity b is higher”

4.       Equal compensation: “If some activities can not be measured, unequal compensation will lead to distortions in efforts.”

Especially in communities of practice, I think there is a fair bit that a community manager can do to make participation in the community a strong signal of expertise and community integration.  I know a few people at Perl Monks who have been able to venture into technical publishing, in part because of their "resume" at the site.

This is fairly speculative, but as Google is one of the major tools I use when looking at new potential hires, I would make the following recommendations:

- have some sort of number/percentile associated with the user so that it can be seen quickly if they are a major contributor to the site

- if the users posts are rated, make it very easy to find their most "acclaimed" contributions to the community

- allow (or better encourage) users to post resume-style content such as education, background, expertise, and contact information.  For a while I knew several people on perlmonks who were basically using the site as their homepage for consulting.

- allow users to control content on their profiles to look professional, this might mean some features (like a graffiti wall, or profile pic) should be optional

This may not be applicable in every community situation, but at least in the fairly commercial domain of software developers, I think there is something to be gained for both the community and the users by making it easy for users to signal expertise, their craft, and professional credentials within the site itself.

--------------------------------------------------------------

oostendo@umich.edu

--------------------------------------------------------------

The main design goal of this reading is to reveal the information that is not observable. Two alternatives to reach the goal: one is singling, and the other is screening.

Singling: Agents (employees) who owns the private information can use education to convey how productive they are.

Screening: Principles (employers) who don’t have the private information provide various choices and the choice can reveal employee’s private information.

Something I observed:

• Employees and employers have different goals. Employers care about outcome, but employees care about wages.

• The distention among agents should be obvious. For example, in the singling case, high-productive workers will be better off by attending higher education, and low-productive will not.

I've given some thought to signaling and screening in e-communities, and come to the conclusion that most do not require signaling/screening at the front gate.  I think that signaling and screening are absent in many online communities for two reasons.  First, many online ventures are funded by ad revenue (or in my community, subscriptions), which returns based on community size, not quality (the assumed goal of signaling/screening).  Second, and probably more importantly, the anonymity of the web makes signaling and screening that much more difficult.  It's difficult to know whether Joe Newbie is really named "Joe" at all, let alone whether he has excellent knowledge of nuclear science or a black belt in tae kwon do.  The methods for signaling over the web just aren't as cost-effective and robust as the "real world".  (Although there are companies working on this problem... Google 'rapleaf' if you're interested.)

While many e-communities are not strictly policing the front gate, a lot of them differentiate users after they've been involved within the community.  Subcommunities have formed within my community that screen based on various criteria, such as character level or longevity of the account holder, among other things.  These criteria indicate that the community member is of high value... has a willingness to work within the game (level) and a dedication to the community (longevity).  This screening within subcommunities is not hampered in the same ways as screening at the community level:  self-organized subcommunities benefit more from member quality than quantity, and within the context of the game there is no anonymity... you are known by the name and accomplishments of your character.

Signaling is a means by which information owners reveal abilities and aptitude to others. Screening is a means by which uninformed parties classify informed parties. Conditions for signaling and screening must favor accurate self-assessment to ensure credibility. For the informed party, the costs of inaccurate representation must outweigh the benefits. Otherwise, low productivity employees have an incentive to misrepresent themselves as high-productivity employees.

Two examples of signaling:
1) HD-DVD players often include many "free" movies to lure prospective buyers away from Blu-Ray players to signal product value. HD-DVDs are cheaper to produce than Blu-Ray discs and content may be more economical for consumers.
2) LinkedIn provides members with the option to vouch for each other. Recommendations from a former employer will signal an employee's potential value to a prospective employer.

Two examples of screening:
1) Standardized tests like the SAT and GRE can be used by uninformed parties (higher ed.) to categorize informed parties (applicants) by intelligence and create barriers to entry.
2) HotEnough.org is a dating site that screens members for attractiveness. Prospective members have to send in a photo and existing members must rate them 8 or higher on a hotness scale to be offered membership. Monthly fee: $10.

I think popular open source projects have good signaling mechanisms. Becoming a developer of any respected open source project requires signaling to indicate expertise and commitment.

• New developers have to spend a lot of time and effort to make a high quality code change as soon as they join. Alternate lower cost route is to start with bug-fixing and code reviews to signal their understanding of the project. This way they can climb ranks before becoming a core developer of the team.
• For expert programmers there is only a low cost to make a quality code change right away. So they can easily signal their expertise and join ranks directly.

As pointed by Nate, allowing users to build their profile - ranking, etc is a good screening mechanism. Only users who have the commitment and expertise to spend effort will contribute to the site. A user is less likely to make a high quality post unless he/she knows the benefit to making quality posts.
The next design alternative I have is 'to have stricter boundaries for sub communities'. This is related to Powazek's idea that all communities should have barriers to entry. Example: if posts in a particular sub-forum are screened by moderators for relevance and quality new users will reveal their type so that moderators will have a lower bar while screening their posts.
The goal of both signaling and screening is to enforce entry barriers to community so that existing members can have a credible identity.

Signaling model is to introduce signal: education to
differentiate different type of workers in the job market. Simply, different
type of workers will choose different action under the same purpose of
maximizing their utility. In the basic
model, education is a pure cost, however, in practice, education and production
is positive related.

Screening principle is widely applied in explaining
phenomena in organizations, for example, why companies pay more to more
experienced workers? How a company set price for bundles of products? It
provides a way to achieve self-selection under information asymmetry.

Applying these economic models to the design of online
communities, I think screening principle is also a guideline to explain why
online communities treat older users and new users differently.

IF I understood this correctly, with a bit of help from
John(GSI), it is in the best interest of any company to stratify its employees
via signaling and screening.  Without
these mechanisms the company wouldn’t really be able to differentiate between
the high type and the low type.  The low
type would easily try to pass itself off as the high type time and time
again.  Furthermore, these stratification
agents allow a company to place value on these respective groups ($20 or
$50).  Anyone, if I got this wrong please
correct me.

Here is what the Wiki,wiki says:

SCREENING

For
purposes of screening, asymmetric information cases assume two economic agents--let's call them Abel and
Cain--where Abel knows more about himself than Cain knows about Abel. The
agents are attempting to engage in some sort of transaction,
often involving a long-term relationship, though that qualifier is not
necessary. The "screener" (the one with less information, in this
case, Cain) attempts to rectify this asymmetry by learning as much as he can
about Abel.

The
actual screening process depends on the nature of the scenario, but is usually
closely connected with the future relationship.

Examples of Screening:

• Banks will often
  screen people interested in borrowing money in order to weed out those who
  won't be able to pay it back. Banks might ask potential borrowers for
  their financial history, job
  security, reason for borrowing, assets, education, experience
  and so on.
• Firms
  interview
  processes are a method of screening, using the conversation to learn about
  the person's personality (by way of mannerisms, attitude and dress) as
  well as other factors.
• The dating and flirting
  that goes on every day is a form of screening, where people use a large
  variety of cues to avoid undesirable mates.

SIGNALING.  

SIGNALING
IN THIS READING=
EDUCATION

Spence
discovered that even if education did not contribute anything to an employee's
productivity, good employees would still buy more education in order to signal
their higher productivity to employers. (Economists sometimes call this the signalling hypothesis in education,
often cited as a reason why government should not subsidize higher education
for workers: more education allows workers to be paid a higher wage but does
not make society more productive.) Bad workers, for their part, would accept a
lower wage rather than pay the higher price (for them) of getting more
education.

Example
of a simple Signaling model:

In the job market, potential employees seek to sell
their services to employers for some wage, or price. Generally, employers are willing to pay higher wages to
employ better workers. Since employers are not always able to observe potential
employees skills and productivity, they use education as a way to estimate the
abilities of potential employees

I'll try to explain how this reading relates back to eCommunities (hopefully I understand it myself).

Users can signal their intents to a community though things like their user name and profile picture and information. It might be easy from a glance to see that a certain user might fit well in a community if he displays signals special knowledge about a subject. For example, a user could use the handle "longcat" and be accepted into a weird web meme forum. Or something.

Communities themselves might screen by requiring some kind of specialized knowledge. My favorite example is this captcha used for signing up to the quantum random bit calculator. 

I think that this website and its point feature is by far the best example for this theory. People are signalling by writing hopefully relevant and interesting comments. The authority in form of Paul is screening for ... hmm ... well for whatever criteria he is exactly screening for ;)  Based on this screening he is distributing benefits in form of grades. Some people only have to write one witty entry and get a lot of points from their fellow students and others have to write a whole bunch of comments to signal by number of comments instead of number of points.

What do you think is that more or less accurate, or am I totally on the wrong track?

There have been quite a few posts on good examples of signaling
and the importance of signaling to get the “right” kind of members for a
community. Let’s look at it in the light of the topic of this week: barriers to
entry from the website or communities point of view. The dilemma is that while
every website wants to have the “good” type of visitors- there are only so many
“good” type of visitors and their attention and involvement is the scarce
resource.

Building an initial barrier to ensure that only people who
put in an effort to cross that barrier get in may not be the best way to go
about it – particularly if a community is just starting out. For one,
communities are in a way experience goods. For users who are unsure whether to
join a community or it is important to provide them an easy way to explore the experience
of being a part of the community. This is in line with Milogram-Roberts statement
that private information responsible for adverse selection may be discovered by
spending effort and attention - but it has associated costs which may make it
unattractive to use.

Therefore it might be more practical to allow for easy
exploration but to build in some kind of screening measures to allow agents to
signal their quality by their contributions to the community once they are in.
For example, in my community PagalGuy.com moderators are chosen from amongst
people who’ve known to have spent on average 10 hours a day on the site and
provided useful advice thus signaling their quality. On the other hand to view
the forums you do not even need to register, but registering allows you to talk
to others, message them, post.

There are also ways we can leverage other signals as an
indicator of the quality of a person. A common example is the beta signups for
websites.  Before companies usually have
resources to accommodate large amount of users they want to try their product
with a smaller number of users. For example initially facebook only allowed
people to log in with University email ids. Even know if I want to join the Microsoft
network I need to have a Microsoft email id. Effectively, Facebook is
leveraging a real world signal – a Microsoft email id – as a network to screen
people who join a particular group in Facebook. Gmail ids initially could only
be got through referrals (if I remember correctly you got only 6 invites!). The
idea was to use real world social networks to limit the invites to potential
good types and not spammers or others.  

The difficult part of this metaphor isn’t the signaling but what is the wage earned from the signal? I think that an easy way to view this article is that the employer is the site admin. Signaling to the admin will gain you admittance to the site. But, I think that many other people are assuming that the employer is the community. To me it makes more sense to view the community as the entity that we gain the wage from, and the wage in this example is attention. In many communities the users signal to each other in order to gain status in that community. This suggests that the wage that they receive is the attention of the community, or their accolades.

signaling:
uses a criteria of differentiation that is correlated (acts as a proxy for) some invisible quality that various parties wish were visible

screening:
provides a variety of options targeted at sets of self-selecting choosers

not sure what more there is to say about that :-)

--------------------------------------------------------
PHartzog@umich.edu
--------------------------------------------------------
The Universe is made up of stories, not atoms.
--Muriel Rukeyser

It is nice to see many interesting posts here. In my opinion, the implication of  signaling and screening in our context is going further than just entering the community, for example, becoming a member of a club. It would actually involve any kinds of interaction among agents inside and between the communities. As the most principle economic theory, it implies various kinds of interactions in people's social life, in any kind of transaction people are doing. For example, how would two people using online dating website interact, under a very typical situation of private information.
Thus, the boundaries resulting from these asymmetry or incomplete information circumstances, not only lies when people enter a community, but more on the many levels of people's participations in the community (or say entrance of particular participation status on the community). For example, how one would sell his idea about Obama is more credible than others on a forum, thus gaining the public opinion leadership on the site. On the other hand, people often screen via these signals, when they usually have no sufficient information. Interesting thing is that, on a real circumstance, things are not as simple or unique as education versus productivity. Because in reality, people use many different kinds of methods for both signaling and screening, which can be diverse, useful, weird, individual, biased.....etc. In addition, different from the traditional communities for example like a company club, which is suffering less on information incomplete, on the online setting, this becomes more serious and people have more options for both singaling and screening. For example, people can manipulate signaling by change the photo and others can use data statistics for screening measures. I guess those are interests to explore too.

As I was reading this, I started wondering if there are many situations in which various community sites are able to reduce the need to stratify their members through screening. The use of various user-contributed content sites to source media may be one such example. 

In the past, when a company needed a certain image (and they were not going to take/create it themselves), they may not have turned to their employees, but they probably had to turn to some limited set of stock photography catalogs or companies that handled the screening for the company, in a sense, picking images and with which photographers to work. As these catalogs moved online, it was possible for the company to search more catalogs at a lower cost (and less screening of which photography companies to use was required). As user-contributed content sites grew, it became possible for companies to search and license images from amateurs and professionals in the same way. When a company looks for an image this way, no one seems to be screening the photographers, but the decision is based on the work products (images) themselves.