Misclassified Master Requests Reclassification

This has occurred to me during other discussions. I think a good addition to the "trust-metric" calculation would be that the person being certified has the ability to set an upper limit on his/her own certification level. After all, who knows my limitations better than me. Mechanically, this could work by letting the trust metric calculate my level, and without input from me, fall as it may. But if I certify myself lower, then *that* would be the level I am rated at. If later, I felt I had risen to the next level, then I could certify myself as such (or just decertify myself) and let the trust metric have total control again.

In other words, I should have the simple ability to successfully certify myself lower than the metric might certify me (but not higher).

current hacked-up mod_virgule is capable of supporting any certification "areas", specified as certs/*.xml.

. for example, i created skills.xml as a test:

<?xml version="1.0"?>
<skills>
  <date>2000-04-16 06:09:53</date>
  <info name="Open Source Developer Skills" default="None"/>
  <seeds>
    <seed name="lkcl"/>
    <seed name="anton"/>
  </seeds>
  <levels>
    <cert name="None" levelc="#800000" darkc="#006060" cname="red"/>
    <cert name="VBscript Expert" levelc="#c0c0c0" darkc="#606060"
cname="grey"/>    <cert name="Programmer" levelc="#c0ffc8"
darkc="#008000" cname="green"/>
    <cert name="Hacker" levelc="#c0d0ff" darkc="#2040ff" cname="blue"/>
    <cert name="Guru" levelc="#e0d0ff" darkc="#8000c0" cname="violet"/>
    <cert name="God Of Code" levelc="#e0d0c0" darkc="#804020"
cname="violet"/>
  </levels>
</skills>
 
~            
~
~
"sample_db/certs/skills.xml" [readonly] 18L, 681C  

sorry, people, misunderstood. scottyo's post explained what i suspected but didn't quite think that that was actually what lilo was saying.

ok.

*why* would you not want to accept trust given to you?

it seems to me that you can refuse to accept responsibility or simply not _use_ your privileges, but short of contacting everyone and saying, "hi, please don't trust me"...

lilo, i think that the most trustworthy people are the ones that *don't* want the responsibility.

I think scottyo has the perfect solution to this one. FWIW, if I could make my self-cert take effect, I would.

lkcl asks "*why* would you not want to accept trust given to you?"

Well a master cert does not currently imply any more trust than a journeyer cert, except it allows you in turn to certify other masters. I don't think this is a trust issue.

I like the current specification of the trust metric (in light of recent discussions, I may be almost alone in this ;-), and part of that specification says that "a Master writes clearly about the work and its broader context, and serves as a mentor to others in the free software community." I don't do that, so I don't think I deserve a master cert. It's that simple.

joey, thanks for being honest: then you should probably contact the individuals who certified you as "Master" and tell them that their assessement of you is wrong: so probably should everyone else [do likewise]. nothing wrong with that. appreciate lilo pointing it out.

Hm, contacting all of them would be a lot of work. I'm lazy and prefer technical fixes, which is why I like scottyo's idea so much. (And I appreciate a compliment as much as the next guy.)

OTOH, joey, a lot of people have seen the work you do and are obviously impressed by it. Maybe if you don't yet see yourself fitting all of the master requirements, you could consider this an encouragement or a challenge to take the next step and try to fulfill the rest of the requirements.

I think the same goes for other uncertain masters, or even uncertain journeyers and apprentices. If your peer ranking seems a little inaccurate in your eyes, there are two ways to fix it: 1) some new advogato implementation so you can lower yourself OR 2) take a look at the things that you don't do and do them!

This is implicit in the definition of "apprentice", so why not apply the same thing to "master"? no one becomes a master overnight. :)

I like scottyo's idea that an individual should be able to reduce his or her certification. At least, I think I do. :) There are clearly situations where the group will miscertify you, with the best of intentions, and it seems reasonable to allow the individual to decide what level to accept, up to that provided by the group.

On the other hand, what's the maximum level an individual should be able to certify? If everyone else certifies me as Master, should I be able to certify others as Master if I only consider myself a Journeyer? Or should I be limited to the level I accept as my proper level? Probably the latter, but I'm not sure.

Luke, as usual, you have Provided Code, which is a Good Thing. :) I agree that there should be different skill areas which can be certified individually. I'm not sure it is always the case that you should only be able to certify others in your exact skill area. I think it depends on the skill area. For example, a successful Master Technical Manager who is only a Journeyer Coder may should be able to certify Master Coders I think. Hmm, that requires thought.

Anyway, thanks all for the interesting comments so far.

I don't believe that it is constructive to brand the free software communities as composed mostly of apprentice collaborators.

In fact, never mind `constructive'. I don't even believe it's correct. We're talking about groups of people who can put together complex software like compilers, operating systems, kernels, desktop environments and industrial-strength web and mail servers, and can do this despite multiple timezones, cultures, nationalities, and native languages, for the most part employing low-bandwidth text-based channels like mailing lists and IRC. If these are apprentice collaborators, what does that make the average proprietary software company who still has a problem with letting people telecommute?

The proof of the pudding is in the eating. Let's recognise that what we're looking at here is pretty damned impressive, and let's start examining why it does work. The assumption that it doesn't is not suported by the evidence and a slight on the people involved.

On 22 April 2000, dan wrote:

I don't believe that it is constructive to brand the free software communities as composed mostly of apprentice collaborators. In fact, never mind `constructive'. I don't even believe it's correct. We're talking about groups of people who can put together complex software like compilers, operating systems, kernels, desktop environments and industrial-strength web and mail servers, and can do this despite multiple timezones, cultures, nationalities, and native languages, for the most part employing low-bandwidth text-based channels like mailing lists and IRC.

Most successful open source contributors are Apprentice collaborationists.

Most serious open source contributors are Apprentice collaborationists.

If these are apprentice collaborators, what does that make the average proprietary software company who still has a problem with letting people telecommute?

The proof of the pudding is in the eating. Let's recognise that what we're looking at here is pretty damned impressive, and let's start examining why it does work. The assumption that it doesn't is not suported by the evidence and a slight on the people involved.

My point in defining levels above Apprentice, which I consider a significant achievement, is to suggest that greater collaborative skill has significant utility. It's not to suggest that our project people are amateurs or unskilled.

Thanks again for your comments.

I agree with lilo on this one and I've taken the time to change some of my classifications from yourneyer to apprentice.

Also, maybe it's not completely right that every person get's the maximum rating that gets through the trust metric. Eg. if 3 masters certify someone as apprentice and 1 master certifies that person as a master, then I guess that person should be either apprentice or yourneyer, but not master like what would be currently possible...

if there is more than one path from a to b (e.g a->b and a->z->b) and you take the weight of the *minimum*-weighted path: the entire graph tends towards zero.

not exactly what you want.

the only sensible action is to take the maximum weight, given several alternative paths. in this case, that's trust metrics.

On 23 April 2000, riel wrote:

I agree with lilo on this one and I've taken the time to change some of my classifications from yourneyer to apprentice.

Also, maybe it's not completely right that every person get's the maximum rating that gets through the trust metric. Eg. if 3 masters certify someone as apprentice and 1 master certifies that person as a master, then I guess that person should be either apprentice or yourneyer, but not master like what would be currently possible...

Personally, I feel that all this classification smacks of elitism. One should not rank based on particular skills, everyone has different strengths and this is what makes us unique individuals. If you really do want to rank someone, rank them based on respect. Respect can be taken to mean many things such as I respect Linus for both his coding skills and sense of humor. It also nicely promotes people that have made useful contributions to the community in a "collaborative" fashion. This serves to eliminate all the various classifications that people are coming up with. If you choose to cert based on varying skill sets, it is a never ending thing as one can continually come up with new classifications.

lilo has made some interesting points although his diary entry regarding phil serves to lower my respect for him somewhat as a claimed "skilled" collaborator. Coming back to respect based ranking, I think a more accurate word may be repute and hence by extension, reputation. In this community, reputation is everything and as stated in various articles before a major although not necessarily the only driving force in the open-source/free software world. Realistically, you *will* feel pretty good about yourself if something you created is widely used by others and praised as a GoodThing (tm).

So how does someone get ranked based on reputation/respect?

Let's see. Person A and B both each write a large piece of software that is widely used, however Person B is also an asshole (you know who you are) and is abusive towards people that he/she considers to be of inferior intellect. One can reasonably be expected to rate Person B lower (especially if you have been on the recieving end before). By extending this further, you should be able to rapidly be able to build up a reasonably accurate picture of people in the community. I believe this will lead to a better way to rate people's contributions which ultimately shows how much has been given back to the community.

I also think that the ranking system should be changed to a scale based one i.e. 1 - 10. The various titles appear to have caused more harm than good and does not provide for enough granularity. Of course all of this ranking business is purely subjective which is why I think it is fair to rank based on reputation and reputation alone.

I will have to think more on the exact details on how to implement this, suggestions are welcome and hopefully a patch to follow (if I ever find the time).

On 23 April 2000, jauderho wrote:

lilo has made some interesting points although his diary entry regarding phil serves to lower my respect for him somewhat as a claimed "skilled" collaborator.

A simple change to mod_virgule could be to use self-evaluation to allow you to certify yourself at a lower level than the one given by the other advogatos, in case you consider that you've been over-evaluated.

This would be a way support modesty in software.

ok, ok, all right: hint taken, i'll do it. btw i really need reviewers and feedback. mod_virgule.tgz.

I regret that lilo has felt the need to retreat from advogato. I thought he was at least one among you that I might actually learn something from.

I'm not sure why he left, but it's too bad that anyone has to feel "driven away." I may be wrong but I suspect it has something to do with jauderho's comments above. I went to jauderho's page and read his diary entry (dated April 23, 2000 as I write this) and noticed the following statement.

I am withdrawing all my certs of advogato folk and feel free to withdraw your certs of me. I know what your skills are and those who have certified me know my skills (at least I hope you do or you probably should not have certified me in the first place).

At least lilo has proven himself (at least in my opinion) to have more character than jauderho. Where jauderho boasts that he doesn't want to participate, but does, lilo has silently withdrawn. I hope you both reconsider.

Note that jauderho works for KPMG, which is the majority owner of LinuxCare, which is where lilo works. Some of the other parties in this little snitwar work for the Puffin Group, which was acquired by LinuxCare some time ago. I'd have to guess that this little unpleasantness is the result of nasty corporate politics spilling over into an unrelated forum.

Shame on all of you.

My error. KPCB, not KPMG, is majority owner of LinuxCare. I suspect there's a commonality of ownership between KPCB and KPMG, however.

Heh. Here is my reply to both kelly and scottyo... I certainly do not think that lilo was driven off simply by my pointing out what I view as inconsistent behaviour, I do not have THAT much influence. Like he says, some people are not meant to work together. If you reread my log again, you will see that I have met lilo in person before and while I like him as a person, I am under no requirement to agree with him, that's why freedom of speech is good.

I withdrew my certs because I do not believe in the current certification system and while it is not strictly a form of protest, I am not able in good faith use something I do not believe in. I do believe in ranking systems but what I am looking for is a representation of the "organic" recognition that is assigned to people in the community. However, it may be that a tangible representation is impossible because eventually ego gets involved and then bad things happen. I am interested and supportive of the article and diary portions of Advogato. They allow me to catch up with people in the community that I may otherwise not have the time to communicate with.

Let me state for the record: "KPMG has no 'commonality of ownership' between it and KPCB nor does KPMG have anything to do with Linuxcare". In fact, I once interviewed to work for LC and lilo was one of the people I interviewed with and a major factor in them giving me an offer. I ended up not taking the offer but I am grateful for the opportunity. Likewise I am not able to speak for the internal politics of LC but rest assured this is not some corporate spat. I hope this clears things up a little.

I haven't seen anything in any of the comment threads here or in anyone's diary that I think merited lilo's advogato suicide. Was there associated conversation in some secret forum that drove him to it?

I just wanted to correct Kelly on the KPMG-Linuxcare-Puffin relationship. Linuxcare acquired the Puffin Group last year. KPMG has nothing to do with Linuxcare, except that they are a client. Let's not all look for consiracies where none exist, especially in simple factual errors on Kelly's part.

Oh, and KPCB is not a majority owner of Linuxcare either... They're an investor.

On 23 April 2000, riel wrote:

Also, maybe it's not completely right that every person get's the maximum rating that gets through the trust metric. Eg. if 3 masters certify someone as apprentice and 1 master certifies that person as a master, then I guess that person should be either apprentice or yourneyer, but not master like what would be currently possible...

This is why I proposed to use two different methods for evaluating trust and skills. Trust (of identity and integrity) works well with a maximum rating: if 3 trusted users are not quite sure that your account really belongs to you but 1 trusted user (who knows you personally) certifies your account at the highest level, then I think that it is correct for your account to get the maximum rating. This is the formula that is used today. But this does not work well for rating your skills or reputation, because the maximum rating is not a good way to evaluate the mixed opinions of the community. That's why I suggested to use a weighted average: if 3 users certify you as Apprentice and 1 user certifes you as a Master, then the simple formula that I proposed would give you an Apprentice or Journeyer level, as you expected. Please read my article for more details...

By the way, I think that all certs/rating systems have some bad sides: eventually, someone will disagree with their own rating and complain about it, like lilo did. But they can be very useful too: you can only know a limited number of people, but the certs/rating systems allow you to see who these people know, and so on. Although there is no point in certifying someone who is already known by everybody (e.g. Alan Cox), it is very interesting to certify someone who is not so well known. This allows others to get a first impression of someone they do not know.

As no doubt the issue was a misremembering of something Kelly and I were talking about, I thought I'd clarify what Kelly and I had discussed.

1) Ted Schlein, a partner at Kleiner Perkins (KPCB), and KPCB themselves each own approximately 21% for a total of 42%.

2) Paul Vais, along with the VC firm he's affiliated with, each own 8.5% for a total of 17%.

Thus, just taking into account the VCs that comprise the Office of the CEO (along with Art Tyde and Pat Lambs), they own (as individuals and company reps) 59% (aka a majority) of Linuxcare.

Nevertheless, Kelly's original point was valid: there seemed to be hositility from several Puffin folk towards lilo, something that, for some of us, seemed to come out of nowhere.

_Deirdre