Lambda the Ultimate

I'd be happy to hear suggestions for other heuristics.

I agree that communicating by example is going to be the way to go for those people who aren't aware of the problem. However, I think that the easiest approach will be to point out the policies in particular cases when necessary, going forward. If we highlight older threads, we'd need to give an explanation of the issues with each one anyway, if the concerns are to be understood by someone who doesn't already understand them. In any case, the offending parties have usually already been notified, and there's no reason to drag specific names into it again.

The real issue of concern, btw, is not so much flamewars, or raising high emotions as Kay has put it, but rather that a number of the forum discussions lately have been of low quality. To use the signal to noise metaphor, although the noise has not been high, the signal has been low. The emphasis of the site is intended to be on the material posted to the home page (a community blog), and discussion of that material, which is supposed to set the tone for the site. The discussion forum has been moving further away from this ideal, which has also affected the discussion of home page material.

Note that the "blog not forum" quote is not from the policies (other than a link to that thread in the credits). If there's something in the policies which seems unclear, please let us know. The intent is that a new member could read the policies and at least get some idea of what is and isn't acceptable. Of course, we don't expect that to work every time, in which case the relevant policies will be pointed out if necessary.

I think the emotions raised is an indication of certain principles not firmly established yet. Nobody will debate today the advantages of using a high level programming language instead of raw assembly; the vast number (vast in billions) of LOC written in high level assembly simply proves that. On the other hand, FP is a shady ground, at best, and being elitist does not help. I think recent events reflect this clearly.

I think the emotions raised is an indication of certain principles not firmly established yet.

That may be true, but this is precisely why we're saying that LtU is not the place to debate such cases — because debate is not LtU's primary mission (see Chris Rathman's comment here), and because such debates are not usually fruitful, tending to degrade the quality of the site.

If someone were to write a high-quality paper or article about such a subject, which consisted of more than mere opinion and was well-researched (or at least did not ignore well-known work), then it would be on-topic to post a link to it and discuss it here. While the resulting discussion might still become heated at times, it would at least be centered on something which is intended to further the state of our knowledge about a subject, and would not be a complete waste of the time and attention of those who come here to gain such knowledge.

In the absence of that, as it points out near the end of the policies page, there are other places to indulge in debates, advocacy, and anti-advocacy.

On the other hand, FP is a shady ground, at best, and being elitist does not help.

No, but being disciplined and insisting on minimum standards of discussion quality will help.

Whether pure FP languages like Haskell or Clean ever come to enjoy a larger audience (or somewhat impure FP PLs like ML and Scheme), I think these PLs will influence languages that are already in widespread use (C/C++, Java, C#, VB, Perl, Python, Ruby, PHP, etc..). For example, though it's improbable that a language like C++ will ever aspire to be pure, the work being done with the Boost library is a good indication that C++ will try to adopt some of the ideas. So even if you never get around to finding a good substitute, I'd be a little more cautious in the assessment on FP.

Whatever ones take on FP as a paradigm, the FP languages are quite important from a PL standpoint that goes far beyond their immediate number of users. Of course, there will continue to be debates about whether pure languages will ever be widely adopted - be it OO languages like Smalltalk and Eiffel, or logic languages like Prolog and Oz, or functional languages like ML or Haskell, ....(substitute any number of paradigms, with any number of languages at this point)... these languages will have an effect on the way we program and think, even if we never move to a new language.

Personally, I think there are a lot of languages that deserve to have a wider audience - naively, I think that real competition makes all languages just a little better. But that may asking too much, given the sociology and history of PLs. Still, those of us who enjoy studying PLs on a professional, theroretical, or hobby basis, like to be able to connect the dots and expand our toolbox.

From an LtU meta-discussion perspective, PLs shouldn't be dismissed simply because they are unpopular. They should be examined for the ideas they contain, and mined for the lessons they teach us. In this context, languages like APL, Snobol, and Epigram have important things to say, in spite of the odds stacked against them. Perhaps it's just me, but I enjoy a multitude of languages, even if I only use a few of them in a professional environment. I find it personally annoying to see discussion that focus exclusively on the questions of popularity or viability.

The proclamation that FP languages are too elitist and overgeneralizations about not being useful is against the spirit of those who think there's more to PLs than mere advocacy. In some senses, I enjoy the enthusiasm that the advocates have for their favorite language. But when they step outside of their own favorite language, they can be utterly predictable and tiresome. So although I find what you say about MVC, OOP and C++ to be meaningful, I find your assertions concerning FP are less than authoratative.

My opinion is that LtU is not about proving which languages or paradigms are the best, or which ones will be successful (however one wishes to gauge success). It is, rather, an attempt to discuss PLs in a more considered fashion. In that spirit, I think it is more productive to consider other viewpoints, rather than to constantly dismiss them as not applicable to solving our immediate problems.

Anyhow, it's just a thought, but I can't see the direction that you are headed in your above response will be productive for either yourself, or for any of the other participants. And it's not really a question of whether FP is good, bad, or indifferent. It's more about how one approaches the question in the first place (or - more succintly - whether one is even asking the right sorts of questions).

Quite.

Perhaps it's just me, but I enjoy a multitude of languages

It's not just you. I think most LtU members share this feeling. Come to think of it, this is a good indication whether LtU is an appropriate venue for you or not.

Regarding the publishing of new ideas: I think that in most cases a write up should appear somewhere else, and only a pointer given. LtU does not host longr documents, of the kind needed to substantiate new claims. So this isn't rally a ban on new ideas, but rather a style issue.

Now to the specific types of posts you mention:

• propose a human factors experiment that might shed light on an area of controversy before someone has conducted and published such a result? - If this comes up in discussion, I don't see a problem. A short suggestion would also be acceptable even without context. A long debate about such a proposal, whose details weren't hashed out elsewhere will tend to be problematic.
• create a posting to request literature pointers related to some formally referenced topic? (e.g. citing what you know about and seeking more of relevance.) - Perfectly acceptable. Notice that if this style of posts become too frequent it too can become problematic for the foucs of the site, but normally this should not be a problem.
• express an opinion that questions the assumptions behind a paper and raises an alternative view to try to learn whether such an approach might have been test elsewhere? - Ok, of course, provided this doesn't turn into an endless debate about assumptions preventing others from discussing the paper
• announce an ongoing research project related to LtU topics and to invite interested LtU members to participate? (e.g. a tasteful one time recruiting pitch to announce the formation of working group on a language related topic to take a discussion offsite until some substantive results are in to be reported back to the community.) - A short post is fine.

I would suggest laying out a semi-formal decision making framework with clear tests

For the record:

LtU isn't a public or state owned organization. The final say so will remain in the hands of those running the site and the community. The policies document is merely an attempt to clarify some issues and help moderation. It is not complete nor formal. Moderation activities will not be open to debate about the wording of the policies etc.

Perhaps I should emphasize again the following from the policies page: Posting here is a privilege, not a right.

I think everyone other than truly new arrivals is already clear on this and I am sure that the whole community is more than content to abide by whatever moderation policy the community leadership adopts regardless of its final wording.

Hence, my choice of the term 'semi-formal' in making my suggestion. The idea was more to have a decision tree or something closer to pseudo-code that someone could follow to get a good sense of the acceptability of what they were thinking of posting and not to propose a quasi legal system with appellate review or external arbitration for LtU. The last thing I would want would be to have to wear my lawyer hat here instead of my programming language researcher hat!

So I was not trying to second guess Anton's language, just to emphasize that anything that reduces ambiguity will go a long way to avoiding future problems and ongoing debates over policy. But whatever happens, Ehud and the moderators are providing an incredibly valuable service to the language design research community so I will abide by and support any decisions they make.

I wonder if the distinction between common law and civil law are useful here. It seems like LtU is governed by the principles of common law including traditions, customs, and precedent. We rely on an ever increasing set of case law (front page posts and commentary) to guide us, as opposed to an expanding corpus of statues and policy pages which proscribe behavior. It seems like that distinction could be an interesting area of study for online communities. Anyone have pointers to interesting papers discussing electronic jurisprudence? If nothing else, in the spirit of providing context and grounded discussion, maybe our new constitution could mention which school of legal philosophy our founding fathers hold dear.

I'm not sure that publishing is to be taken in the formal sense - Journals, Thesis, Conference Papers. Publishing material on the internet is dirt cheap, and can be done fairly quick. So I think the emphasis is on shelling out the ideas on an external link as a beginning point for the discussion - instead of throwing out a raw unpolished idea and having everything be self referential within a thread (or discussed totally in first-person mode).

Item 4(b) says: The detailed argument for such a claim should be posted on your own blog or website, and posted as a link on LtU.

Obviously, if you manage to publish in a research journal, we won't hold it against you...

The word "post" might be more appropriate, or perhaps "post or publish"?

First a quick comment that I put here so that it hopefully will be seen: Unless you are waiting to incorporate the ideas from this discussion, the policy and spirit pages should be linked from the FAQ or have their own sidebar link.

Anton asserted that "It is not a forum for the publishing of new ideas that have not previously been published elsewhere."

or specifically from the (current) Policies page:

LtU is best used to discuss ideas that have been published and argued elsewhere. It is not usually a good place for design discussions, developing ideas from scratch, or debating opinions.

I agree with Peter on this issue. The word "published" is probably too strong, but more to the point I think that LtU can (and has/does) provide a good place for new ideas. Of course, these ideas should not be vague and should be very much in the spirit of LtU. I actually don't disagree with the latter sentence of the quoted section of the policies page: particularly, I agree with the statement that LtU is not a place for design discussions or "persuasive" debate. The "developing ideas from scratch" part is where I'm wavering somewhat. I agree LtU is not for building ideas from whole cloth, and that as mentioned repeatedly, they'd be better handled by being fleshed out elsewhere and then perhaps linked to on LtU, but occasionally papers and the discussions around them spark new ideas that I believe can be fruitfully discussed for everybody even if the result is that the idea isn't that great/new.

This may be partially "covered" under:

We expect that such posts are likely to most often be directed at newer members, or those who haven't previously been active posters. Members who are regular posters are more likely to already be familiar with the site policies, and also are likely to be tolerated more easily if they occasionally stretch the rules.

A statement that I'm quite glad was added and was quite appropriate.

So as an attorney, I would suggest laying out a semi-formal decision making framework with clear tests (bright lines wherever possible and balances of interests with any biases weighted in where needed) and as many exemplars as we can draw from past experience that newer members could more easily follow to assess the quality of their potential postings.

This, even quite attenuated, I strongly disagree with. New posters have an entire siteful of exemplars on LtU style to draw from; they should be lurking and/or reading back before posting. It should be more or less clear what is and and is not acceptable for LtU to new posters. The primary point of the policy page, as I see it, is mainly to clear up what not to do and to give something to point to for community enforcement of the policies. A "decision framework" is completely unnecessary and possibly damaging for community enforcement. The community collectively, but implicitly, knows what it wants and will always know what it wants even as that changes; the community will attempt to maintain that and so far has usually succeeded, though occasionally a "benevolent dictator" is quite handy :).

Unless you are waiting to incorporate the ideas from this discussion...

Yes, that's the reason the links aren't in the FAQ yet.

It seems that the wording regarding new ideas should be reconsidered. I am not sure why the concern, but we can soften the language some and still get the message across, I think.

The word "published" is probably too strong

Well, to me the word "published" includes publishing on the web, seeing as how we're in the midst of a 21st-century revolution in media an' all. But as the policy page says, "terminology has a context", so I'll bow to the retro elements in the audience who still think in terms of dead trees, and clarify that point. ;oP

but occasionally papers and the discussions around them spark new ideas that I believe can be fruitfully discussed for everybody even if the result is that the idea isn't that great/new.

Right, and the policies certainly weren't intended to prevent that.

I think one place where the current policy document falls short is distinguishing explicitly between posting new topics to the forum, vs. posting comments to existing topics. The restrictions around "new ideas" are really intended to apply more to the former than the latter, and as Ehud has pointed out, as much as anything else it's a stylistic issue: LtU is intended to host links to papers and articles, not the material itself. Another aspect is just an attempt to draw the kind of bright line which Peter asked for, to rule out posting to LtU things which would more appropriately be published (oops, I mean posted) on one's own blog.

It's inevitable that ideas, whether old or possibly new, are going to come up in the discussion surrounding a paper. I think the last thing we'd want to do is inhibit that, so we'll clarify that point as well.

The primary point of the policy page, as I see it, is mainly to clear up what not to do and to give something to point to for community enforcement of the policies.

Precisely.

I think one place where the current policy document falls short is distinguishing explicitly between posting new topics to the forum, vs. posting comments to existing topics.

I know of no other blog nor forum on the Web where new comments get posted to the old topics as much as on LtU. I find that fascinating, and I believe the cause of this oddity is that the (best of) forum discussions on Lambda are almost a representation of the forefront of the open research problems. So when a new peace of research appears that's relevant for an existing discussion, a new post gets added to keep the discussion up-to-date.

I really like this phenomenon, both because it's intriguing on the sociological level and for my selfish purposes: A discussion of this sort is a great resource for learning and keeping my references up to date. I'd like it to be encouraged as much as possible. Would it be possible to insert some statement to that effect in the policies? It would also be nice to have a sort of "top 10 list of the ongoing discussions" list somewhere on the site. Such a list would be a good resource for new members, and at the same time it could serve as a quick reference to a particular area of research. Which 10 discussions are the best would probably have to be determined by the contributing editors...

I have in my mind still a few old threads that I'd like to post replies/questions to, but I haven't done the necessary reading + experimentation yet.

But, I don't think it's about the top 10 discussions. Unlike at slashdot, discussions here don't die because the moderators moved on. People do read old topics and respond. So, what we should encourage is well thought out questions or comments, no matter how long it takes to formulate them.

We should also discourage mini-tutorials directed at a specific person; we should be able to assume that we can just point a link to a paper or a book and end the sub-thread if we believe that the person needs to be better informed on something.

I agree that the fact that old threads remain active (made possible largely by the fact that new comments pop to the top of the recent changes page) is a unique and attractive character of LtU. One reason is that old threads remain useful references. However, I am unsure why this should be relevant to the policies discussion.

If this feature of LtU is something we want to keep and nurture, that should be stated in the policies. Minimally, the policy could state something like

Before you post a link to some exciting new research as a forum topic, do spare some time to search the archives. It's possible that (a) the link has already been posted, or (b) there is an existing forum topic discussing previous research in the very same area. In such cases, adding to the existing forum topic would be preferable to creating a new one.

Of course, this would put even more "burden of proof" on new members who'd like to open a new forum topic. I don't consider this consequence of the proposal a benefit, by the way. This could be helped by some technical improvements to the site, but that discussion would really be off topic.

Thanks, concrete proposals like this are great. The only change I might make to this is to say that it can be appropriate to create a new topic if the previous discussion had a different emphasis, in which case including a link to the old topic may be appropriate.

Of course, this would put even more "burden of proof" on new members who'd like to open a new forum topic. I don't consider this consequence of the proposal a benefit, by the way.

In what way? Could you expand on this?

This could be helped by some technical improvements to the site, but that discussion would really be off topic.

I'd be interested to hear at least a summary of what you have in mind.

Thanks, concrete proposals like this are great. The only change I might make to this is to say that it can be appropriate to create a new topic if the previous discussion had a different emphasis, in which case including a link to the old topic may be appropriate.

Thank you. I'd like to make another small amendment: please replace the word "archives" by something else. I just realized that the link "Archives" at the front page, which would be a natural thing to follow for a new member, leads to the archives of the old LtU site.

Of course, this would put even more "burden of proof" on new members who'd like to open a new forum topic. I don't consider this consequence of the proposal a benefit, by the way.

In what way? Could you expand on this?

First to clarify: I wouldn't have proposed this if I thought it was a bad idea overall. I hope it does lead to better-researched postings and more people reading existing topics of relevance. But there's no denying that it would also lead to fewer new forum topics, if for no reason then because searching the archives takes time. I'm guessing Ehud would disagree, but I don't see the recent increase of number of new postings as a problem. Even if the signal/noise ratio is geting lower, I find it preferable to no postings at all. Perhaps I have too much time for reading.

Which leads me to technical improvements. I'd like to be able to search the existing postings more easily, and be confident I didn't miss anything relevant to something I'm about to post. I know some of the old guard would be confident with no need for a search, but what is a new member supposed to do? The only way to search currently is by using keywords, i.e., the technical terminology, but not everybody knows the jargon for what they're interested in. Or they could search for one keyword, and miss a topic that uses a synonym.

I don't really have a solution, but if you want an outline, I think there are roughly two groups of solutions: either classify all topics into a tree or graph of related topics, or formalize the existing keywords approach.

One possible solution in the second group: Require every topic to specify a list of keywords. For links to research papers, the keywords are readily available. Then have a page somewhere on the site that lists all keywords ever used for a topic. If I can see the full list of keywords, I'll know what keywords to use for search.

Given a database containing keywords and topics, of course, it would be an easy matter to generate links to all past topics and postings that used a particular keyword, or to suggest correlated keyords, etc.

That was enough for an overview. And anyway, whatever gets done should go only as far as it helps informing the discussion. LtU should strike a balance between a research reference and pure discussion. We don't need another CiteSeer.

I think changing the search facility is overkill. Using the google search (instead of the Drupal one) almost alwsy wroks (and search is never 100% anyway).

I think the concrete suggestion you made makes sense. Thanks!

I would agree with this subthread except that I believe this function should be taken over primarily by the (suggested) LtU wiki. Of course, it would be no reason not to revive old discussions, but I think it would be more useful to all to distill out the main lines and collect together similar threads into "categories". This would also somewhat cover the "searching" issue.

You are right about the wiki, of course.

A "decision framework" is completely unnecessary and possibly damaging for community enforcement.

In my experience, relying on each new person to inductively build a decision framework from a plethora of examples doesn't work well because of human laziness.

Someone wanting to push a subjectively (in their own mind) brilliant idea, will probably read the FAQ policy doc, may read the long policy doc, and might read a half dozen past postings if you are lucky.

The sort of framework I had in mind would be a one screen flow chart bearing its own sidebar link with each decision point linked to a concise exemplar.

They key is to make it really concise, otherwise you might want to force all newbie posts through a moderation filter to allow time for the inductive process to work its magic.

In my experience, relying on each new person to inductively build a decision framework from a plethora of examples doesn't work well because of human laziness.

1) LtU seems fairly unique, so your experience may be less applicable here.

2) This isn't what I'm suggesting. The part you quoted says nothing about the "newcomer", it is about the community. A newcomer may well not take the time to get a feel for LtU style, but if they violate it you can be sure the community will react, most likely simply by pointing out what they've done wrong. If they disagree they may well leave. If they are persistent then that's when the benevolent dictator comes in. Most things have not even gotten to the second stage.

A newcomer may well not take the time to get a feel for LtU style, but if they violate it you can be sure the community will react, most likely simply by pointing out what they've done wrong. If they disagree they may well leave. If they are persistent then that's when the benevolent dictator comes in.

An assumption that seems to get made quite frequently is that there is an obvious "LtU style" that can be gleaned by reading LtU for some period of time.

I'm not so sure that this is true.

I've been reading LtU since its inception and have been posting on and off for several years now, and I'm STILL not too clear what the boundaries of the LtU style are, or if they even really exist beyond the changing subjective tastes of long-time members and the administrators.

I have my own preferences and opinion about what this style is, of course, but I'm routinely surprised to see someone who, to my reading, is a naive-but-sincere newbie get savaged as a "troll", while someone else who, again to my interpretation, lacks any real desire to understand PLs and PLT more deeply than they already do, is tolerated and conversed with repeatedly and at length. (Not mentioning any names of course ;-) )

Similarly, sometimes otherwise upstanding members of the LtU community engage in threads that are questionable to my eye. A newcomer can't tell that this is tolerated because "we know Joe is normally a good poster" rather than because it is an acceptable thread. If they post in the like vein, they may be surprised by the drubbing they get.

For these reasons, I agree with Peter that a clear statement of principles and examplars would help immensely if we genuinely want new, thoughtful people to join our community. Otherwise, they may come to the conclusion that we are an old boys network with highly subjective and xenophobic standards of savoir faire.

If the latter is actually what we want (nothing wrong with that really), then I think we should drop the pretension of being "newbie-friendly".

It is partly our not wanting to commit to one side or the other that I think leads to confusion and bad feeling.

We think the policies document does that. If you disagree helping out with constructive suggestions might be a good idea.

the changing subjective tastes of long-time member

That's the whole point. If LtU really belongs to the LtU community, this is exactly as it should be.

We think the policies document does that.

I agree that the new policies are a good starting place.

I interpreted Peter as supporting the necessity for such policies, and for refining them to be more specific and concrete as we go. In contrast, I interpreted Derek as questioning the necessity for such a policy refinement, that just reading LtU will impart the necessary knowledge.

helping out with constructive suggestions might be a good idea.

If I conceive of any more refinements to the policies that I think will help, I be sure to speak up.

If LtU really belongs to the LtU community, this is exactly as it should be.

Then maybe what is missing is a clear statement of what is required to be a "member of the LtU community in good standing" whose subjective tastes can be expected to matter. (Become an editor? Read LtU for a long time? Post at least 20 quality posts?)

Perhaps the next step is to find some easy way for newcomers to be able to TELL who the established community members are (based on whatever criteria) versus those who are merely loud-mouths.

This would make it easier for them to recognize whose opinions they should particularly "take a hint" from.

find some easy way for newcomers to be able to TELL who the established community members are

I thought about this several times, but decided it's a red herring. In any community, the only real way to decide who is who is to bo invovled and observant. Now, we can say look at the user id number: the lower it is, the more likely the member is an old timer. But this really isn't a very reliable indicator. Like I said before in this thread I think you don't really need, nor want, formal rules about such things. Indeed, I think they are likely to cause more harm than good. This is why I think the solution might be between what Peter and what Derek said, and that this middle ground is served rather well by the policies proposal, which tries to capture the essence of LtU without providing formal rules.

It might make sense to mention that some members are contributing editors, and that their opinion carries mroe weight, but ultimately in my mind the real criterion is what the community as a whole thinks regarding whether something is problematic, not any one user (and when I decide to pull the plug, I try to reflect what I hear from others, not just my personal judgement). I know that getting a feel for collective opinion can be difficult, but maintaining a sense of community *is* difficult. This is why we are in this process to begin with: the larger the community became the more issues that came up.

Notice that this thread is an example of the process I am talking about: I didn't decide on the policiy (though I am the "oldest" member), nor did I poll the contributing editors, or even delegated the decision to some subset. I opened this topic for discussion.

I interpreted Peter as supporting the necessity for such policies, and for refining them to be more specific and concrete as we go.

"Refining them to be more specific and concrete as we go" sounds fine to me, and I'm sure we'll do that as it becomes possible and/or necessary. For anything more immediate, we're open to concrete contributions.

However, absent any specific critiques or suggestions, it's not clear to me that a whole lot more is needed right now. It's not necessary that every new member instantly absorb the policies, written or unwritten. Among new members, it's only a small minority that need any help integrating. If they get that help, regular members will also have less reason to react in questionable ways.

Regarding whose opinion is expected to matter, the Policy Enforcement section contains the statement "If you, as an LtU regular, are unsure about your objection, it is best to mention it publicly." I think it's fine if "regulars" are self-identifying. Publicly stating such objections helps us all to get a better sense of what kinds of posts are eliciting negative reactions, and if necessary, discuss the issues and perhaps refine the written policies.

Marc, before responding to some specific points, I want to mention that your comments — a couple of which are quoted on the Spirit of LtU page — were a welcome influence on my own thinking about LtU policies. You talked about collegiality, the advantages of non-anonymity, and the issue of over-generalizing from one's own experience, all of which were addressed in some way in the policies document. So, as far as your "own preferences and opinion about what [LtU] style is", I'm very interested to hear anything you have to say about that.

Re your specific points, I've responsed on some fairly minor points below, followed by a response on something I consider a pretty major point, namely newbie-friendliness vs. an old boys network.

Re-responding on one point from a slightly different angle:

I interpreted Peter as supporting the necessity for such policies, and for refining them to be more specific and concrete as we go. In contrast, I interpreted Derek as questioning the necessity for such a policy refinement, that just reading LtU will impart the necessary knowledge.

I think it's quite possible that reading a future LtU, with policies in place, will more easily impart the necessary knowledge. Following policies should reduce the inconsistency of what is and isn't tolerated, and hopefully reduce the provocations which lead to many of the unproductive posts.

I can't say much about Peter's suggestion because it's too unspecific. I can imagine versions of it which would be good, others which would be bad, and plenty in between. Someone will have to do some work to make any of these versions happen. However, no matter what we do, implementing and refining policies will be a process, not an instantaneous transition. "A good starting place", as you put it, is all we really need right now.

I'm routinely surprised to see someone who, to my reading, is a naive-but-sincere newbie get savaged as a "troll", while someone else who, again to my interpretation, lacks any real desire to understand PLs and PLT more deeply than they already do, is tolerated and conversed with repeatedly and at length. (Not mentioning any names of course ;-) )

Similarly, sometimes otherwise upstanding members of the LtU community engage in threads that are questionable to my eye. A newcomer can't tell that this is tolerated because "we know Joe is normally a good poster" rather than because it is an acceptable thread. If they post in the like vein, they may be surprised by the drubbing they get.

Having policies ought to reduce such inconsistencies. But even with policies, it's to be expected that regular members get will more leeway, simply because respect and trust is earned, not automatically granted. That's just a fact of human relations which newer members need to be cognizant of.

For these reasons, I agree with Peter that a clear statement of principles and examplars would help immensely if we genuinely want new, thoughtful people to join our community.

"A clear statement of principles" sounds wonderful, and again, we're open to suggestions and contributions on that. There are a few principles in the current policies. A primary principle, as I see it, is stated in the third paragraph of the Purpose statement: "LtU is foremost a place to learn and exchange ideas". Many of the policies follow from that: does some behavior further the goal of learning and exchanging ideas, or detract from it? If anyone wants to identify or propose other principles, I'd be delighted.

(Of course, as already discussed, the "exchange ideas" bit has some caveats, which I plan to try to clarify.)

Otherwise, they may come to the conclusion that we are an old boys network with highly subjective and xenophobic standards of savoir faire.

If the latter is actually what we want (nothing wrong with that really), then I think we should drop the pretension of being "newbie-friendly".

I trust that I'm not alone in the opinion that this would be a tragedy, comparable to the tragedy of the commons that we're currently trying to steer clear of. It's not clear to me whether this comment implies that you think the policy document could be interpreted in this way, or whether it was mainly a response about the viability of absorbing LtU style purely by osmosis. Regardless, I think the point is worth addressing.

I think that often, the "old boys network" approach is just an easy way to avoid having to deal with integrating new members in a more equitable way. The whole point of the policies are to help with that integration. They're not intended as a pretense. If that's not clear, I'd appreciate any suggestions that would help make it clearer.

I think that the foundations for our "standards of savoir faire" can be found in the Getting Started thread, and if that's perceived as xenophobic or subjective, we should try to address that. Someone who makes a sincere effort to study programming languages — which realistically, must include studying some of the literature on the subject — is typically going to be quite welcome here. I think a number of the problematic threads have been driven by people who have not yet made that effort, and to be fair to them, probably didn't realize that there was any reason to.

I would like to see the site being friendly to sincere newbies, and I think it mostly is that, with a small number of exceptions recently. Those might have been avoided, or more easily handled, if written policies had existed.

If the policies seem indirect or vague in places, it's at least partly because where possible, they attempt to focus on underlying issues that have led to problems, rather than on proscribing specific behaviors, i.e. there was an attempt to address causes rather than symptoms. If that, or anything else, has led to a lack of clarity, then let's try to fix it.

As I said above, implementing the policies will be a process. I think it's more important right now that we have written policies, which LtU members generally approve of and (hopefully!) understand, than that they be complete or perfect.

I can't think of any specific concrete changes I would make in Anton's draft, beyond my gut feeling that some sort of mini-flow-chart to summarize it might be of values if others also felt that such a guide would help. Perhaps some of our newer members could comment on whether any work in that direction is needed or if the policy document itself or in conjunction with this thread has given them enough guidance.

Otherwise, I would commend to your attention an essay by Clay Shirky called,
A Group Is Its Own Worst Enemy that I first read in Joel Spolsky's "The Best Software Writing I". It might add some helpful context to our discussion.

Shirky's article has been mentioned a couple of times in previous discussions. While interesting, I'm not sure what lessons to take from it for the LtU case. Shirky himself summarizes the problem:

So there's this question "What is required to make a large, long-lived online group successful?" and I think I can now answer with some confidence: "It depends." I'm hoping to flesh that answer out a little bit in the next ten years.

In any case, I don't think that the problems with LtU are particularly intractable. It's just that we were previously relying almost completely on self-regulation, with no explicitly agreed-on rules. That makes it difficult for someone playing a moderator role to decide what to moderate, and difficult for people posting to know what not to post.

To start, lest anyone reading this thread mistake my comments for a criticism of the policy or at least a "damning with faint praise" of the policy, I want to make clear that I think the current policy document does an excellent job of expressing what LtU is about.

(I would have to be an even bigger contrarian than I am to argue with a document that links to my own words. ;-) )

My strong preference is for expectations and assumptions to be out in the open, since I think that communities function better that way. The policy accomplishes that goal, and there may be additions that can be made as we go to refine it.

My involvement in this thread is just chearleading for this point of view. ;-)

as far as your "own preferences and opinion about what [LtU] style is", I'm very interested to hear anything you have to say about that.

The role that LtU plays for me has evolved over time, so I'm finding this a very challenging question to answer succinctly and productively.

My own reading over the last few years has made me much more of a theory wonk, and in my heart of hearts I would like LtU to serve as my virtual "faculty lounge" or "PLT conference", since I don't really have access to those venues as an amateur "academic". It does serve these functions to some extent, especially as a source of papers.

At the same time, I recognize that it is probably not practical or reasonable to expect the kinds of changes that would be required to make my "dream LtU", such as possibly higher barriers to entry, and I also appreciate that LtU plays different roles for others (as it has for me at different times).

Things can't be that bad though: I've never managed to stop reading LtU every day. ;-)

This one is problematic... Well focused questions are a good thing, of course, that's not LtU's focus, so putting thins on the policies page might be a bit misleading. Another problem is that questions are often a way to hide strong opinions, which are then lead to the endless debates we are trying to squash (the policy page deals with this issue explicitly).

Thus I am inclined against this suggestion at the moment, but I inderstand the idea Eric proposes. Perhaps we can come up with wording that will make these concerns less likely.

I was a bit annoyed recently when someone made a thread to ask a question, and instead of answering it everyone posted a paper link or two, in some cases to subjects which are only distantly related.

I am all in favor of citations, of course, but sometimes I think it is a way of passing the buck. The best thing is to post a link together with a more concrete explanation of how the information in the link can be applied in extenso, and ideally pros, cons and alternatives, etc.

Lately, also, I haven't seen much code on LtU, which I think is a shame.

PLT is a field which straddles theory and practice; this is an advantage we should try to exploit as much as possible. I think 3/4 of the religious discussions I read could be pre-empted by either bringing more theory or more practice in, at least if all participants are willing (which, granted, is not often the case).

I talked about D-infinity models for a long time before I actually posted that universal type code. And then everybody just 'got' it. I should have done it long before, but I was lazy.

I am all in favor of citations, of course, but sometimes I think it is a way of passing the buck. The best thing is to post a link together with a more concrete explanation of how the information in the link can be applied in extenso, and ideally pros, cons and alternatives, etc.

Frank brings up an issue that I had been thinking about in my previous post, but didn't quite gel for me then.

As a manifestation of my "virtual PLT conference" idea, giving references is often useful, though I can ultimately use Citeseer or Google for that; the thing that makes the difference is the critique and context that the poster can bring to the citation.

Of course, improving those elements would require more of us to feel comfortable putting ourselves on the spot intellectually; how to foster that without risking degeneration is, I would say, the central motivating problem of this thread.

Since several people have mentioned examples, here's a question/answer example that illustrates Frank and Marc's point. I asked the question as well and as clearly as I could, and almost immediately (within an hour or two) had several very clear and useful answers (thanks again, Neel and Derek!) and an interesting discussion to back them up. This was really helpful for me, and I have virtually nowhere to turn other than LtU for this kind of friendly expertise. Obviously I'm biased in this case, but this really stuck in my head, and I hope others agree that this is an example of "good LtU" rather than "bad LtU"...

I think that as an example oriented towards newcomers this is quite scary stuff, so while I agree with the sentiment I think this shouldn't be part of the policies document.

People with agendas hijack threads that deal with completely different subjects simply to voice their vocal and inflamatory opinions, and kill any chance of the original thread to be useful to anyone.

Ehud, this is exactly the sort of pronouncement that can only be understood if we know what specific examples of threads and posts you mean.

There have been a couple posts lately that fit this behaviour exactly in my eyes, and a couple that I could see might be perceived that way by someone else, but which are within bounds to my taste.

Without the specifics, I (and presumably other readers) can't know what the intended reaction should be. Depending on what I interpret the target to be, I might make a radically different choice in what I think and how I behave as a result.

How about you give the examples since you seem to like them and your sentence There have been a couple posts lately that fit this behaviour exactly in my eyes, and a couple that I could see might be perceived that way by someone else, but which are within bounds to my taste is exactly of the type that calls for examples (at least by my standards).

How about you give the examples since you seem to like them

I hope you are joking, and I'm just missing it, Ehud.

You made a claim about what sort of behaviour is acceptable or not at LtU, and you want me to find the examples to explain what you meant?

If you randomly chose a few threads from the history of LtU, I could probably find SOMETHING to which I could apply your criticism in some way or another, and I still wouldn't know what YOU meant.

If a particular post offends me enough, I will respond directly to that post, in the thread in question, and explain exactly what offends me about it. I may equally choose to ignore it if it doesn't pass my threshold; I'm bound not to like everything that goes on in my community, but I'm not going to "call the cops" everytime something bugs me.

According to our new policy document and this thread, we expect people to back up their opinions with some context and supporting evidence. If the principle that LtU is by and for its community is to be more than rhetoric, this must apply to administrative pronouncements and meta-discussion as well.

Ehud, there is one clarification I would like to make that might help you get where I'm coming from.

I have always had a very difficult time telling when you are speaking with your admin hat and when with your member hat.

Upon reflection, I realize that my default setting for you is "admin". Unless you signal otherwise (explicitly or by context) that's how I read you. (Others have this too?)

When I challenged you to come up with examples, it is because I perceived you as making an administrative pronouncement. If I had interpreted you as speaking in member mode, I would probably NOT have asked for examples for the very reasons Anton brings up.

I have noticed that Anton usually marks his administrative posts with [Admin] to make his hats clear, and as a result, I never seem to confuse his roles.

Perhaps, if you do the same, I can clear the default "admin bit" for you, and cut you some more slack. ;-)

Thanks!

My concern about giving negative examples involving current or past threads is that we unnecessarily single out individuals who, at the time, were posting in an environment with almost no written policies, and very little enforcement.

If we're going to discuss adding a "no thread hijacking" policy, then I think it's sufficient that we agree on the principle. Examples can follow as they arise, once the policies are officially in place. In that case, Marc might flag some posts as too OT, if he feels it's warranted, Ehud might flag others, and Anton might have to be flagged a few times (Voltaire? Sheesh! I blame Paul. :)

If that results in disagreements on how a policy should be applied, then we should discuss it. Until then, I think we should accept that the exact parameters for applying policies are a work in progress (actually, a work that hasn't even really begun).

My concern about giving negative examples involving current or past threads is that we unnecessarily single out individuals who, at the time, were posting in an environment with almost no written policies, and very little enforcement.

I don't want to unduly stigmatize individuals either, and as I say, if I have a problem with something I try to say it directly to the person in question (or as directly as you can in a public medium).

A policy like "no hijacking" is just too darn vague, though, and some digressions are as stimulating (or even moreso) than the original thread. (I enjoyed the "Leibniz/Voltaire" dialog, BTW ;-) )

To stick with my "faculty club" example, if two colleagues have rambled on a little too far off topic you have a couple choices: to ignore them and have your own conversation, or say "Hey guys, let's go back to a topic we all like."

Obviously, this breaks down in an environment not necessary made up of respectful peers with shared background knowledge.

I guess it all depends on who we think the LtU target audience is, and as I've suggested earlier in this thread, some of the problems we may have coming up with a workable policy might revolve around different ideas about who this audience is/should be.