Creating Our Own Peer-to-Peer Expertise-Finder

I‘ve been waiting for Google, which has already provided a definitive ‘know-what’ information finder and ‘know-where’ place finder, to follow up with the definitive ‘know-who’ people-finder. My initial thought was that only Google and one or two other giants could get enough profile with this to get everyone to participate and accept it as the standard, and hence achieve the critical mass to succeed where so many Social Networking tools that have tried to do this have failed.

But then it occurred to me that there is a profound difference between ‘know-what’ and ‘know-where’ on the one hand, and ‘know-who’ on the other: Finding the former are complicated search problems; finding the latter is a complex problem. Google can write an algorithm to point you to the documents most likely to be useful to you on subject x, and they can create maps to point you to location y. You don’t have to do anything but ask. And although the numbers are vast, there are only a finite number of documents and places on the planet.

By contrast, any meaningful people-finder would require active and regular participation of many people. Assessments of expertise are too subjective and change too quickly over time for any kind of algorithm to ‘interpret’ from data that are already ‘out there’.

But any kind of top-down, managed expertise-finder will inevitably fall victim to the same problems that have afflicted Linked-In and other social networking apps: Not broad enough participation, data that is stale and which no one is motivated to keep current, and the tendency of people to try to ‘game’ the system to portray themselves as more popular and expert than they really are.

The only thing that will work, I believe, is a Peer-to-Peer solution, one that works with existing ubiquitous tools and which makes it easy for anyone, regardless of what platform they are working on, to participate with little or no incremental effort. When addressing any complex problem, we need to give the solution the opportunity to evolve as the result of the collective intelligence of everyone.

That is an imposing challenge but not an impossible one. What we need to do first is develop a high-level spec for a system that no one will build. The spec will be merely the initial set of principles and guidelines that will influence how we participate. The ‘crowd’ will tell us if some of those principles and guidelines are wrong, and what’s missing, and we’ll change them to reflect that wisdom and imagination.

Here’s my first cut at some of those principles and guidelines. We need the people who know the Internet best, both as a technical and social phenomenon, to add to this list — we won’t get it ‘right’ the first time, but the closer we get the list in the early stages (or, to use complexity terminology, the more valuable our initial set of attractors and barriers), the faster something useful will start to emerge from it.

  1. The terms for expertise should be folksonomic, not taxonomic: It is futile to try to design a taxonomy of expertise — there are too many terms and types and they change too fast. Let everyone decide on their own terms to define their, and others’, expertise, and let the system accommodate them. This is like what does with its music tags — it doesn’t set out a predefined set of genres, it allows its millions of users to self-define tags that mean something to them, and lets the ‘crowd’ settle the matter. A critical corollary to this principle is that the terms are not hierarchical: Your terms for expertise can be as general or as specific as you want. Some people may be looking for generalists, others for highly specialized people — the system doesn’t discriminate.
  2. People can define their own expertise: The best initial set of terms of expertise is probably self-defined: We all know ourselves (or think we do) better than we know others. We don’t get to ‘vote’ on our own expertise, but the best way to initially populate the folksonomy of expertise terms is to get a few million people to ‘tell us about yourself’, to define their personal genius in their own terms.
  3. Voting on others’ expertise should be simple: Forget 1-10 scales. The best gauge of expertise, one that is independent of the financial wealth of the voter, is how much time you would be willing to spend listening to and learning from that other person on that subject. Your personal time and attention is the ultimate investment, and your willingness to invest time and attention is hence the ultimate measure of another’s expertise. So the vote for others’ expertise should be as simple as I would be willing to invest (a) lots of time [H], (b) a little time [M], (c) no time [L] with this person on this subject. There is no ‘default’. You don’t vote on another person’s expertise unless and until you know them well enough to intelligently answer the ‘I would be willing to invest time’ question.
  4. Your votes would sit and be maintained on your own hard drive: No submissions to central repositories — we’ve seen again and again that that mechanism just doesn’t work. gets its data by automatically harvesting data from participants’ iTunes as they play songs on their PC — no need to ‘tell’ the system what you like. We need a place that everyone with a PC has in common, and my suggestion would be the Address Book. Add to, or requisition, a field from the address book to use as the expertise assessment field. So if I think John Smith is a terrific expert in cultural anthropology, an interesting guy to chat with on innovation, but less than useless in his self-proclaimed area of expertise, social networking, I’d enter those three folksonomy terms in the expertise assessment field of his e-mail record on my hard drive followed by an H, M, and L respectively. I don’t know whether address books do, or can be made to, date-stamp when these assessments are made, reconfirmed and changed, but the date of this assessment is also pertinent and needs to be captured somehow.
  5. Different expertise ‘scores’ are needed for different purposes: Suppose I’m looking for an expert on ‘knowledge management’ in the ‘health care’ industry. I may want to know who has been rated ‘H’ by the absolute highest number of people in both those areas, or rated ‘H’ in a single tag ‘health care knowledge management’. I may instead want to know who has the highest median rating (2H+M)/(H+M+L) in these areas. I may want to know either of these things but counting only assessments made or reconfirmed in the last month. There is no single ‘score’ that meets all needs, so any scoring algorithm needs to accommodate these different needs.
  6. We need to be able to filter and analyze expertise assessments in many ways: We may only want to see experts who live in certain geographic areas or who speak certain languages. We may only want to see experts who will give us some time or expertise free of charge, or whose rate is less than, say, $50/hour. And we may well want to see the identities of the people making the assessments, and discount those in large organizations who rate everyone else in their organization ‘H’ on everything to ‘game’ the system. The system may in fact evolve to allow us to assign a trust/credibility rating to different assessors, and filter out assessments from those we don’t trust.
  7. We should leave the tabulations to those who do them best: Rather than trying to come up with our own tabulation system, we should simply charge Google, Yahoo, Technorati and other companies that are already expert in search and ranking algorithms with the task. As the tabulators of ‘know-what’ and ‘know-where’ information they have a vested interest in tabulating ‘know-who’ information as well. They can also grapple with the security issues (e.g. accessing people’s Address Books to harvest, or canvas just-in-time, the assessment data to respond to ‘know-who’ search requests).

I can see this evolving in interesting ways. Corporations will initially want to use this within their Intranet firewalls to find experts within their own organizations, and won’t want that data accessible outside the firewall. But information is always trying to be free, and once smaller organizations ‘let it out’, and buyers start looking for and expecting to see their preferred suppliers’ experts’ names showing up on ‘know-who’ search results, the big professional firms will have no choice but to open up the data to the world and let buyers start putting together their own cross-organizational teams of experts.

I also think that being acknowledged as an expert is a double-edged sword, and such a system will start to create genuine ‘markets’ for expertise. People acknowledged as experts who are bombarded with requests for their expertise, and who cannot afford (or do not want) to spend their whole life sharing what they know free, will naturally start to put in personal, market-driven hourly rates for their expertise, and hence filter out most of the requests.

Who knows, some of us might find that we’re acknowledged as experts by more people than we think, and we might even be able to make a living simply on the strength of this system’s ‘word of mouth’.

It sounds very complex and unmanageable, I know, and it is, which is probably why it hasn’t happened already. But there is a clear need for a viable, simple, reliable, easy-to-maintain expertise finder, and once a few million people agree to start maintaining the information that would drive it, I think it could explode quite quickly, and evolve just as quickly to meet this need extremely well. The key is not to try to design a centrally-managed app for it, but rather to let it grow and become what it will become, virally and organically.

I’m going to pass my thoughts along to Doc and David W. for a start (since I recognize their expertise in this area as ‘H’). If you think this is a useful avenue for exploration, please talk it up and tell me what you, and others youtalk to, think.

This entry was posted in Using Weblogs and Technology. Bookmark the permalink.

8 Responses to Creating Our Own Peer-to-Peer Expertise-Finder

  1. medaille says:

    I’ve done some thinking on this issue, and while I don’t consider myself an expert at computer programming or software design, I do have a good feel for what I would like to see in this arena. The biggest thing that I see differently then what I think you are seeing, is that I think the (expert finder, the group collaboration tools, the personal desktop [all of which you’ve talked about]), as well as some other things such as a media exchanger (like for local news) and other similarly defined things (directory service, like craigslist) should all be lumped into one tool.There are a couple of reasons as to why this should be lumped into one tool rather than having several stand alone tools. The first and most important is that there needs to be a method of managing interactions between people. In my mind, the current method of specific tools being made for specific needs is very chaotic. I generally consider myself to be a year and a half behind the progressive edge of computer technology, so if I find it annoying to have to continually search out new tools so I can find like minds then other less technologically-able people probably just completely ignore the bulk of it and thus would be missing out as well as the network would be missing out on them. To me, it makes sense that the people and how they group themselves by the result they desire to obtain should be the first priority and everything should branch from that, rather than the tools defining the group structure. I would be fine if it was as simple as something like myspace, but with some tweaks to fit in tightly with the rest of the system. There needs to be one place for all the people to be grouped to make the network more able.Another reason is that it just makes sense that once you have all your connections able to be usable to have access to being able to do something with them, without them having to download a certain interface that they aren’t amiliar with just to work and discuss with you. As an example, I wouldn’t want to have to teach someone how to install and use skype just to communicate with them, and I certainly wouldn’t want to try to manage 18 different tools in that manner just to work with them.For the most part, almost all items are defined in similar manners. There are only so many common methods of information exchange. You only have 2 of the five senses to worry about, time interactions, spatial occupancy, user modification, and publication type to worry about and they can be combined in different ways. To start out with, you could choose to only exchange words (in audio or text) and/or images (video, 2d documents), etc. I know there are some things that get outside of that box, but mostly, tools stay in there. This means that a lot of tools are redundant under the hood, despite differences in the surface of the way they interact with it. It makes sense to me to have some sort of framework to house everything. Kind of like how mozilla has extensions.On a similar note, the means of sorting something are similar and google isn’t providing a completely adequate solution yet. The biggest drawback is the lack of locality being integrated into all networking tools. While it is nice to have the full internet at your disposal, its sheer girth makes it difficult to get in touch with people you can physically interact with if you so desire. Social interaction/collaboration tools should have a map integrated into them in order to easily visualize and sort experts etc.——————————–This is some half formulated concept in my head, and thus I can visualize it better than I can explain it, so its probably sounding fairly scattered. I do have some general thoughts on this whole issue in its broad context. All users need to be have some sort of privacy/encryption built in. They need to be assured that they are free to exchange information in whatever way they would like. It’s of no use if the FBI, CIA, RIAA,or any other group/person can start intimidating people based on what they are sharing. Freedom of information is a strength of the network and the government recinding/infringing on those rights should not be tolerated if it can be helped.Decentralization vs. Centralization. This is something that isn’t so black and white. From Illich (Tools of Conviviality), we know that there are two watersheds as we progress from diverse systems to industrialized systems. I think the goal, obviously is to stay between the two watersheds. Right now the internet is too decentralized and its productivity is hindered. At some point in time, the limitations of using a thoroughly centralized tool start to outweigh the benefits. The goal is to create tools that are flexible enough to empower people but not to limit them. On the other hand, centralization in proper doses does increase productivity. Nature teaches us that there is reliability, safety, and value in diversification. Natures timescale isn’t exactly our friend though as evolution is a relatively slow process and when you force human hands into the creation process things can quickly get disordered like entropy (for both better and worse) if there’s not some sort of central guiding hand to try to keep order from disorder. Our tools need to self-regulate and keep order and prevent disorder. That’s a given. It is more time efficient to achieve the social change we desire, by preventing tools from creating excessive disorder.What is the main purpose of peer-to-peer collaboration tools? The main purpose is to remove power from where it is centralized and send it back to all the people equally. That’s the whole point of blogs, indymedia, etc. Corporate controlled information exchange has stopped being beneficial and is being used in a manner that is artificially limiting what people CAN be exposed to. In all aspects of life, we need to be reclaiming out power and enhancing our power. Proper information exchange allows people to be empowered which is the complete antithesis of what they are now. It is a crucial building block to sustainability.

  2. David Laing says:

    Hi, I understand what you are saying, and the vast issues associated with it. On the other hand, it could be argued that Squidoo ( is an attempt to provide one part of the answer. On the other hand, the existence of a squidoo lens doesn’t guarantee that the lens owner is the appropriate expert, but they _could_ be the facilitator – the person who can facilitate access to that expertise.Just a thought…David

  3. Mike says:

    Have you looked at FOAF?

  4. Jon Husband says:

    Do you remembver something called XpertWeb, developed by Britt Blaser ? britt is a friend of doc’s, as well. I think the technology is just sitting there, and it was / is fundamentally peer – to -peer.Here’s the URL, with some information and a whole bunch of flow charts / diagrams …

  5. Jon Husband says:

    Here’s the first few words that greet you on the site … I think this is more or less what you are wondering about / looking for:Welcome to XpertwebJoy’s Law: “The best expert for your most important project …… doesn’t work in your company.”- Bill Joy, Co-Founder, Sun MicroSystemsAn open source co-op to connect you directly with the best experts available.

  6. The best way to accomplish the voting would not be by having people actually rate how much attention they THINK they would give to someone. Rather, devise a way to measure how much attention someone actually gets. People are terrible at estimating how much they WOULD do. It takes enough effort just to vote that they would tend to only vote when they would either invest a lot of time or no time.I had an idea several years back that was remarkably similar to but it had am insurmountable flaw. I wanted people to rate the URLs. Then the system would let you find people with tastes similar to yours and you could get a reading for whether a URL was likely to be interesting to you.The problem was the effort involved in rating URLs. The genious behind is the realization that by giving people the “service” of saving their bookmarks in a central location where they could access them from any machine that happened to be handy, they got the rating system as a side effect. If something is cool enough that a lot of people bookmark it, then there is your rating. Then the power users discovered that you could find people that linked to stuff you liked and browse their public bookmarks and find other stuff you liked. Viola! The system I was looking for with minimal effort. Folks are falling all over themselves to rank the links for you because you are supplying them with something they want at the same time.

  7. Tony Eric says:

    Email us via our company email account( ( phonesNokia N91 220$Nokia 6060 145$Nokia 6111 132$Nokia 6270 125$Nokia 6270 115$Nokia 6280 125$Nokia N90 115$Nokia N70 115$Nokia Vertu 200$ ( 5140i 125$ ( 6230i 115$Nokia 8800 200$Nokia 6021 105$Nokia 6030 115$Nokia 6680 125$Nokia 6681 120$Nokia 6101 110$Zenith P42W22B 42 Inch Plasma TV Blowout $700Gateway 42″ Plasma TV 16 : 9 Aspect Ratio, Supports1280 $800Samsung HP-R4252 Plasma TV & Monitors $900Samsung HP-R5052 50″ Plasma TV $1,500Panasonic TH-37PWD8UK Plasma TV $650Gateway 42″ Plasma TV 16 : 9 Aspect Ratio, Supports 1280 $700Sony Wega Ke-42M1 42-Inch EDTV Plasma TV $800Panasonic TH-42PWD8UK Plasma TV & Monitors $650Gateway 50″ HD Plasma TV 16 : 9 Aspect Ratio $1,500Panasonic TH-42PWD8UK 42″ Plasma TV – Exceptional 4000:1 $700Panasonic TH-42PD50U 42″ Diagonal Plasma TV Special Order $750Samsung HPR5052 50 High Definition Plasma TV with $1,100Panasonic TH-42PWD8UK Plasma TV $800Philips 42 Inch Plasma TV and Monitor 42PF9630A $1,050Sony VAIO V505DC2 Series Notebook Intel Pentium4-M 2.2 $800Sony VAIO PCG-V505AC Notebook Computer Intel Pentium 4 M $700New Sony VAIO PCG-V505DC21 P4 2.2ghz Laptop Notebook PC $700Sony VAIO VGN-U70 Notebook Computer – World’S Smallest $1,000Dell Latitude Laptop $300Dell Latitude D800 15″ Notebook/Laptop Computer PC $1,200Dell XPS M170 2GHZ 1GB 80GB 256M 7800GTX Laptop Computer $1,150Dell Laptop M60-HQCP741 $750Dell Laptop M60-JJ50G51 $650DEC640 – Dell Latitude C640 Notebook/Laptop Computer $400IBM ThinkPad T30 236697U $1,150IBM Thinkpad T23 26475NU $1,050IBM 2379DKU Notebook Computer $535IBM / Lenovo – N100 Intel Centrino Duo Mobile 1.6G $500Hewlett Packard PG402UA#ABA HP Compaq nw8000 Pentium-M $1,530Hewlett Packard DV131U#ABA HP Compaq tr3000,Pentium-III $1,300HP Compaq tc4200,Pentium-M 750 (1.86GHz),12.1 inch XGA $800HP nc6400 Notebook PC $750Hewlett Packard PR127UA#ABA HP Compaq nc6120,Pentium-m $820HP nc6400 Notebook PC $700HP Pavilion HP DV5250us 15.4″ Widescreen Laptop w/1.83 $750Sony VAIO Pentium M 1.6GHz Centrino Wireless B+G $650Sony VAIO VAIO SZ140P10 Notebook $1,050Sony VAIO VGN-A690 17 Notebook PC $530Sony VAIO Pentium M 2.26GHz Notebook PC (Open Box) $900Sony VAIO S170 Intel Pentium M 1.5GHz, 13.3in WXGA, 40GB $800(

  8. z landolt says:

    The whole idea sounds like a good idea to me. Thanks for sharing.

Comments are closed.