Why projects fail

[Sassy Romano]

Read and spot the similarities.

http://www.hks.harvard.edu/m-rcbg/ethiopia/Publications/Top%2010%20Reasons%20Why%20Systems%20Projects%20Fail.pdf

[mikka Luik]

Ah the memories Not a lot to add except been there and very glad I'm now retired from that particular field =^^=

[Pamela Galli]

A veritable blueprint of what the Commerce Team and LL have done.

[Spica Inventor]

Might I add....not enough software employees....investment money allocation lacking....carelessness....deliberate sabotage. hehe ;-)

[WADE1 Jya]

thanks for posting this report, i hope Lindens look at it too :matte-motes-stress:

what do companies typically do when the big megaproject has obviously failed? just keep working on it for years pretending its still going somewhere? do you think its likely they'll come out & admit it utterly failed completely?

*sighs* wants Lindens to come out of hiding to admit they messed up & MP is over... sooner we all move on the better.

my condolences on Linden's loss.... :catsad:

[JohnMiddlefield]

LL will never come clean with what really happened to the MP. If they did, it would be discovered that it was really the tots and toddlers of the LL staff in the company day care facility that run MP. Ridiculous you say? But look at the results!

[PeterCanessa Oh]

Disappointly this paper read like something from 1985, not 2005.  In particular it assumed "projects" were all of one, monolithic, kind (which is ok if it was implicitly aimed at a specific audience but since it doesn't make that explicit I don't know).

I object to "Create the project plan by working backwards from a drop-dead system completion date" as a way to guarantee failure.  If you can't redefine the deadline you can always redefine "success".  In other words, you can probably do something in the time allowed.  This is at the heart of most RAD/agile methods, represented by the "timebox" concept.

Perhaps oddly for a development manager I've also always been very, very wary of a requirement for "a technical lead who has built a similar system".  If you are looking for technical design to lead requirements that's the way to go - and goes straight to "The IT Solution", if not outright failure.  Requirements are the absolute god in any project and Business Analysts their priests.  It should never be a technical role to understand the business, only to translate the requirements into deliverables.  A technical lead who can't go from well specified requirements to a complete technical design breakdown is not simply inexperienced in that field; (s)he is simply not fit to be a technical lead in ANY field.  Using such a person in that role guarantees they can ONLY give you what they did before and the only way to succeed is DESPITE them, not because of them.

[Kampu Oyen]

Interesting.

So, if someone at LL actually wanted a project to fail, this checklist would have been available the whole time?

[Zanara Zenovka]

7. Three months before the system goes live, assign one junior developer to handle the data migration.

8. Skip the testing phase because the project is way behind schedule.

9. Change the system to support critical new requirements discovered during final development.

10. Buy a commercial, off-the-shelf package and customize it … a lot.

 

They forgot 11. If you're fortunate enough to have end-user input on system requirements, ignore it.

[JeanneAnne]

---

Sassy Romano wrote:

Read and spot the similarities.

http://www.hks.harvard.edu/m-rcbg/ethiopia/Publications/Top%2010%20Reasons%20Why%20Systems%20Projects%20Fail.pdf

---

>>..the failure rate of large projects is reported as being between 50%-80%. Because of the natural human

tendency to hide bad news, the real statistic may be even higher.<<

Failure rate is 100% .. given enuf time..

Jeanne

[Josh Susanto]

I'd like people here to consider what the 10-part list can tell us mathematically about what is probably really happening inside LL.

The Polyanna Pangloss model of behavior analysis I have seen mostly applied on this forum is essentially as follows:

Negative behavior: probably unintentional

Positibe behavior: probably intentional

As what is observable has often tended to put me in a position to disagree with people here about what meanings to assign to Linden behavior, it's somewhat understandable that most of you likely imagine I'm using this kind of an unber-cynical model, which would easily explain why we tend to disagree:

Positive behavior: probably unintentional

Negative behavior: probably intentional

So please let me clarify what model I'm actually using. It's essentially the same model used by practically every person who does professional modeling of behavior with statistics.

Random behavior: probably unintentional

Nonrandom behavior: probably intentional

So how does this relate to the 10-part list?

It relates in terms of understanding whether or not LL's compliance or noncompliance with the essence of some number of the suggestions on the list is better understood as random or nonrandom; that is: probably intentional or probably unintentional.

Assuming LL would toss a coin on each of the 10 items to decide whether or not to comply with it in essence, the outcome we would expect would be 5 positive complies with a standard deviation of 2.5 more complies or 2.5 fewer complies. 

If they would comply in essence with anywhere within 3 to 7 of the suggestions, we could characterize that as indistinguishable from random. It would still mean that they don't know how to make good decisions, but it would not give us any reason to prefer to think that either positive or negative outcomes are being deliberately produced.

That's the easy part to understand. 

The next point I'm going to make is a little bit more abstract, but I think anyone who has read this far can probably grasp it without too much extra effort.

The important point is:

Either a compliance score of 2 or 8 would have equal statistical implications in terms of intentionality, as would 1 compared to 9 or 0 compared to 10.

In other words: One set of numbers suggests intentional positive outcomes and the other set of numbers suggests intentional negative outcomes, but the degree of intentionality statistically implied is equal.

This is important to understand because, if LL complied with only 2 of the suggestions, I think most of you people would be quick to want to forgive them these 2 and credit them with at least doing 8 other things correctly. That is: you would assume positive intention and dismiss as silly the idea that they'd merely got 8 things right by accident.

But if they complied with 8 of the suggestions, how many of you would be equally eager to assume negative intention? What we all know I should expect from you people in such a scenario is exactly what I'm seeing now: people continuing to stumble all over themselves to come up with some remaining way to believe "they obviously didn't really mean to do that".

Please consider again, carefully what I have said about the significance of with how many items on the list LL has complied in principle, and tell me how I might be wrong. 

That done, please tell me:

With how many of the items do you perceive LL to have complied in principle?

If the number is 8 or more, then, from the simple standpoint of behavior analysis, we should say that it is likelier than not that at least one person at LL in charge of one of the decisions has decided with the intention to produce a negative outcome such as is described in the article.

This is not a very diplomatic way to look at the Lindens, I know.

But seeing how much all of you have accomplished with your diplomacy, don't you think it's time that you started to put diplomatic decisions aside long enough to first make objective behavioral assessments? 

 

[JeanneAnne]

Very interesting analysis Josh !! Thanks ...

Trouble with it, tho is:

Positive vs negative is a normative judgement. + or - for whom, under what premises? Perhaps LL is losing $$ w/ SL & has tasked some functionary with killing SL off so that the corporation can devote more resources to other projects, thereby increasing their bottom line. Yer gonna hav2 polarize your +/- continuum by defining "positive" & "negative" outcomes 1st.

Also, not every1 is goin2 agree that those 10 behaviors listed are desirable. Perhaps the Lindens don't think some, or alluv em, are .. & thereby have no intention of conforming to them. How do you know that LL isnt intentionally violating some of all of the listed behaviors because they don't agree with them?

A sample size of 10 is too small to distinguish random from intentional behavior w/ very much statistical confidence. > 7/10 conformity may be suggestive but even unity could still all too well be a random outcome.

So ... you asked >>Please consider again, carefully what I have said about the significance of with how many items on the list LL has complied in principle, and tell me how I might be wrong.<< I have done so.

I don't hold LL in very high regard & have no interest in being "diplomatic" about calling them on their stupidity &/or misfeasance. I agree with the desirabilty of making "objective behavioral assessments" too. All I'm saying is that your analysis doesn't hold water due to small sample size & the arbitrary & subjective nature of the assessment criteria.

Jeanne

[Josh Susanto]

Please see the article in question.

The suggestions are not about how to make a project succeed, but about how to make a project fail.

I've been using the term "negative" here to mean something that should make it less likely rather than more likely that the stated purpose of the project will be met.

In context, this means "negative" is a Marketplace less useful to merchants and customers and "positive" means a Marketplace more useful to merchants and customers.

Whether LL, or, rather, everyone in LL sees the positive and the negative in this way is one of the points I'm trying to address with the analysis. I have not flipped the terms partly to avoid greater confusion, and partly out of respect to any Linden who actually wants to provide better service to merchants and customers. 

As to the question of statistical confidence, I'm not claiming a very high degree of conclusiveness; I'm just trying to dispel the notion that the data is in any way more supportive of the idea that the Lindens simply don't know what they're doing or that they're somehow actually trying to make these kinds of decisions in a way that produces a favorable result for merchants. 

The claim that there is probably a malefactor, by any degree of conficence, only needs to be stronger than the claim that there probably isn't a malefactor in order to be a point worth making. If it's wrong to use numbers to support the malefactor hypothesis, it can only be more wrong to continue to ignore the numbers as a way of trying to support a counterhypothesis of no malefactor; a counterhypothesis which meets with a lower burden of evidence only for social reasons, and not for scientific ones. 

But let's put some of my more charitable understatements aside for a moment...

Not having all the hard facts myself, I think there are merchants on this forum who are still inclined to disagree with me about the probable motives of the least benevolent Linden, and who would yet concede that it looks like LL followed all 10 steps to making a project fail (please think about this for a few seconds).

Assuming that the Lindens literally have no idea how to manage a project like the replacement of Xstreet with anything comparable (or better), we should see behavior equivalent of tossing 10 coins in the hope of getting 10 heads, and yet somehow getting 10 tails. This is an outcome we should see on one project out of about every 2 to the power of 10, or 1 in 1024. If all 10 steps were followed, this puts Linden about 511 projects ahead of schedule in terms of producing such an unlikely result. 

Even if we could believe this kind of bad luck to be possible once, how are we to reconcile it in combination with the additional displays of what I guess we are to assume is other astronomically bad luck in terms of the selection of code deployment dates, somewhat expounded upon here: http://tribes.tribe.net/secondlifemarketplaceproductsearch/thread/9354a5f2-05b0-4fbe-b898-5a8dea6fd770 ?

An additional point that bears making here is that the list of 10 items is not a list of 10 things that must be done in order to make a project fail. It is a list of 10 things ANY of which can cause a project to fail. 

For a project to succeed in spite of one, or even more than one of these suggestions being complied-with in principle, is certainly not impossible. But it's exactly the kind of gamble that hiring people with any training or experience at all is intended to completely avoid; it's the whole reason why we don't just toss coins to make project design decisions.

For a team to produce a predominance of such bad decisions that exceeds what we should project from a random process (3 to 7 at a confidence interval exceeding 51%) indicates a team of people about whom the kindest thing we can say is that to use them to make decisions might be no worse than simply flipping a coin.

So, assuming your own intuitions are correct, that this team is performing at the chance level (despite their good intentions), wouldn't it be cheaper for LL just to hire one person to flip a coin for a few minutes whenever a decision needs to be made about the direction of the Marketplace?

That we are led to believe by LL, itself, that chance operations are not driving decisions regarding the marketplace should less encourage us to think that they all want the Marketplace to offer improving service to merchants and customers than it should encourage us to think just the opposite. 

So, while I'm stretching the meaning of the available data, those who make claims opposite to mine are trying to stretch the meaning of data that doesn't even exist.

Who is more wrong?

 

[JeanneAnne]

>>As to the question of statistical confidence, I'm not claiming a very high degree of conclusiveness;<<

Ok, just so long as were clear on this ..

>>The claim that there is probably a malefactor, by any degree of conficence, only needs to be stronger than the claim that there probably isn't a malefactor in order to be a point worth making.<<

Null hypothesis is that theres no malefactor .. Design an objective non-arbitrary test of this hypothesis & thereby refute it w/ p <.05 that your results arent due to chance .. & then you'll be on2 something !! oh yeah .. & do so in sucha way that the assumptions of parametric stats arent violated .. or .... use nonparametric analysis & thereby loose statistical power .. Meanwhile, its mere speculation that malefeasance is @ play rather than mundane incompetence. The converse is also true, of course .. but mere opinion couched as objective analysis or hypothesis testing is still mere opinion.

Jeanne

[Toysoldier Thor]

---

JeanneAnne wrote:

>>As to the question of statistical confidence, I'm not claiming a very high degree of conclusiveness;<<

Ok, just so long as were clear on this ..

>>The claim that there is probably a malefactor, by any degree of conficence, only needs to be stronger than the claim that there probably isn't a malefactor in order to be a point worth making.<<

Null hypothesis is that theres no malefactor .. Design an objective non-arbitrary test of this hypothesis & thereby refute it w/ p <.05 that your results arent due to chance .. &

then

you'll be on2 something !! oh yeah .. & do so in sucha way that the assumptions of parametric stats arent violated .. or .... use nonparametric analysis & thereby loose statistical power .. Meanwhile, its mere speculation that malefeasance is @ play rather than mundane incompetence. The converse is also true, of course .. but mere opinion couched as objective analysis or hypothesis testing is still mere opinion.

Jeanne

---

/ME scrambles onto Google "definition" to learn what a "Malefactor" and "malefeasence" is so /ME can figure out what is being discussed here.

mal·e·fac·tor  (m**Only uploaded images may be used in postings**://img.tfd.com/hm/GIF/schwa.gif" border="0" align="absbottom" />r)

n.

1. One that has committed a crime; a criminal.

2. An evildoer.

ohhhhhh

[Josh Susanto]

>Null hypothesis is that theres no malefactor 

A secondary and equally strong null hypothesis being that there is also no benefactor.

But how many people are just as willing to accept that?

The shape of the data says that if people will accept that there is no one working to damage the Marketplace, then people should also accept that there is no one working to improve the marketplace.

Moreover, if we can take benefactors for granted in spite of the data, we should also take malefactors for granted.

Moreover, the total utility of the Marketplace has declined rather than improved steadily over many months, so if we should assign any positive or negative disposition to anyone acting nonrandomly at LL, we should assign negative more readily than positive.

For every improvement in out-world commerce utility since the very inception of the Marketplace parallel to Xstreet, a larger utility reduction can be paired, and then some. 

For how many more months would this have to be true before it's reasonable to conclude that, by simply not leaving things as they are, but instead changing them repeatedly for the worse, someone at LL must be intending such a result for some reason?

>Design an objective non-arbitrary test of this hypothesis

Oh, I have done that, certainly.

But posting the results of that gets one banned, and the posts erased.

Thus, I'm trying to stick to what data is being allowed. 

[JeanneAnne]

>>A secondary and equally strong null hypothesis being that there is also no benefactor.

But how many people are just as willing to accept that?<<

Yes, agreed.

I think that people expect that a corporation & its employees operate so as to benefit the corporate bottom line. Doing so may or may not benefit merchants trying to sell things on the browser MP. One might like to assume that by benefitting merchants & buyers, the interests of corporate profit are concurrently served, but this isn't necessarily so. LL's business model may be construed so that LL's interests & the interests of merchants conflict. This may be intentional or unintentional. There may also be factions w/in LL serving their own interests by seeking to reduce corporate profit in order to drive share prices down. Or there may just be simple incompetence & inattention at play.

>>The shape of the data says that if people will accept that there is no one working to damage the Marketplace, then people should also accept that there is no one working to improve the marketplace.

Moreover, if we can take benefactors for granted in spite of the data, we should also take malefactors for granted.<<

I don't think that we can take benefactors or malefactors for granted. In the 1st place you must define who is being benefitted or harmed by the actions of these agents. The corporations? Merchants who sell on marketplace? Buyers? The RL economy? The interests of one or more of these groups may be benefitted while those of others are being harmed. The same person may be benefactor to some & malefactor to others. Parsimony would suggest that no active benefactors or malefactors exist; that whatever benefit or harm accrues is simply the outcome of LL employees doing their jobs however competently or otherwise.

>>Moreover, the total utility of the Marketplace has declined rather than improved steadily over many months, so if we should assign any positive or negative disposition to anyone acting nonrandomly at LL, we should assign negative more readily than positive.<<

Well, I like MP because I can't log inworld from work but I can access Firefox & MP thru it. I have had trouble with MP orders all along so I'm not sure if it's gotten better or worse. It's always been pretty bad in my experience. Because of how bad MP is I don't order anything expensive from it. Only freebies &  very inexpensive items have I purchased via MP. The real virtue of MP, to my mind, is that it allows me to see what's available & to find things I'm looking for, to be purchased inworld later on from home. Sometimes, tho, a merchant will have many more things available on MP than they have on display in their inworld store. I guess this is to save prims on the sim or to save on rent or tier for the store. In any case, the few expensive items I've purchased I found on MP, then either purchased them directly from an inworld store or contacted the vendor & had them meet me inworld in order to demonstrate the item & answer any questions I may have had about it. I've found vendors to be very helpful in this way, and also helpful when inexpensive items I've ordered from MP weren't delivered.

>>For how many more months would this have to be true before it's reasonable to conclude that, by simply not leaving things as they are, but instead changing them repeatedly for the worse, someone at LL must be intending such a result for some reason?<<

"... for some reason..." Like I said above, the reason may be because interests other than merchants' are being served. Or there may be no reason. Despite your claim to have objectively tested the null hypothesis of no benefactor/malefactor, the truth is that it would be very difficult to refute or substantiate such a hypothesis - to tease out active benefit/harm from the results of actions taken at random. We are talking human intent here, the motive behind actions taken by LL employees. Social scientists may have protocols for testing hypothesis regarding motivations but I'm way skeptical of any such methodologies. In fact, I don't even regard the social 'sciences' as being experimental sciences at all. Why do Lindens do what they do? I dunno .. Just cuz .. seems to me like.

>>But posting the results of that gets one banned, and the posts erased.<<

I know. The censorship of these fora sucks. Actually, I'd attribute more malfeasance to the "moderation" of these fora than I would to any perceived attempt to undermine the MP for whatever sinister reasons. Seems to me like there's an active effort being made to run off the most interesting & flamboyant posters.

Jeanne

[Josh Susanto]

>Yes, agreed.

Thanks, I appreciate that. 

I'm more interested in getting people to consider the argument than in getting them to say they agree, though, so thanks for that even more.

I'm using terms like "benefactor" and "malefector" here strictly in terms of the question of the intention to provide greater or lesser utility to merchants, and because someone else offered the term "malefactor" a few months ago, which I thought was helpful in distinguishing between the idea of a conspiracy and the possibility that one person with something like a compuslive personality disorder, someone with a convoluted grudge, someone being influenced from LL, or even someone who is just very confused about some basic methodological  concept could potentially do enough harm to merchant utility to offset all the best efforts of the other Lindens. 

It certainly occurs to me that the whole team at LL could be fanatically bent to the task of making the Marketplace useless, but I think it's just a lot less unlikely that the bad behavior of one person in a position to influence others would be the real explanation.

>I don't think that we can take benefactors or malefactors for granted. 

Right. That's why I use words like "if". I do not ask people to take the presence of a malefactor for granted. I only ask that they accept it as no less consistent with observable behavior than the opposite idea that someone is actually trying to fix the marketplace.

I'm seeing some very telling changes in the last couple of days in terms of the way LL is interacting with merchants, and I hope that will continue regardless of whether or not anyone has left the company or moved to a different position within it. I just need to mention that before I go back to discussing the pattern of failure, so that it's clear I'm not talking about the most recent activity.

In terms of the pattern of failure, though...

Practically the only thing I can see LL's merchant defenders citing in support of their claim that LL is trying to make the marketplace more useful to merchants is the repeated suggestion of this by LL. I don't think this has any evidentiary value as support for the benevolence angle because if LL were trying to make the marketplace gradually useless without saying so (for legal reasons, for example), they would quite likely tell us the same thing; that they are trying to fix it. 

So one easy thing to do would be to put aside anything LL has had to say about whether they're trying to fix the marketplace or to make it useless, and just look at their actions. That's where the 10 part list comes in. Forgetting what the stated intentions are, anything approaching total compliance with the 10 steps to failure should more signal an intention to fail than an intention to succeed.

And that could be random, sure. 

But if there is neither benevolent not malevolent intention behind any of the failure decisions, then why bother to tell merchants either thing? That LL tells merchants they are trying to fix the marketplace surely must mean at leas something; either that that's what they intend to do, or just the opposite.

>Well, I like MP because I can't log inworld from work but I can access Firefox & MP thru it.

I think that's a major benefit to a lot of merchants. There will always be continued thinly-veiled classist rumblings, from merchant sim owners, especially, that this allows a lot of "undeserving" people the opportunity to make money without being willing to commit more completely to participation in the in-world experience, and the costs that that "should" entail.

There's an assumption by quite a few people, stated or unstated, that out-world commerce is detrimental to in-world commerce. I think there's some truth to that, but I also think it's a very simplistic way of looking at the whole SL economy as a dynamic entity. Some people say that SL doesn't need the marketplace. I think that the reverse concept is probably at least equally close to the truth; that if in-world shops were all shut down and those merchants had to use the marketplace, they would do so, but that there would just be even less damend for land. It seems to me that if there's too much land for the number of people using it, killing the marketplace just means even fewer use options for new users, so, while it might force people to use more land in the short term, it would probably be worse in the long term by encouraging fewer total new users through reduced commerce options. But I could be wrong. If someone is trying to cripple the marketplace in order to force commerce to go in-world and increase demand for land, that might even be the very formula that LL needs to survive. I don't have the numbers on that. It just sound like something that someone at LL could think will work without having any more solid a reason to think it than I have to think what I think.

So is the marketplace being gradually crippled?

I can't say that nothing at all has improved, but I think if you look at the total balance of system utility from the end of Xstreet to the present, you'll find that Xstreet was a slightly better platform than what we have now, and that not all of the problems are attributable strictly to the transition from Xstreet to Marketplace; that many things, such as the corruption of the 14xxxxx cluster, are more recent. 

>"... for some reason..." Like I said above, the reason may be because interests other than merchants' are being served.

I think that that meets the situational definition of "malefaction". Not in terms of proving that someone at LL likes to throw Gideon Bibles into pools at motels, but just in terms of doing harm to the interests of merchants, especially while pretending to try to do just the opposite.

>Or there may be no reason

Yes, that's possible. But there being no reason for something is somewhat inconsistent with the habit of constantly giving reasons for it. 

>- to tease out active benefit/harm from the results of actions taken at random.

I should explain that I meant to treat the question of benefit/harm essentially in terms of producing a set of results that we should expect merchants to prefer as contrasted with a set of results over which we should expect merchants not to prefer by comparison. I suppose that the 10-part list could also be read as a list of steps to producing negative merchant affect as subjectively reported by merchants. I guess we could continue to do a lot more to refine my terminology to make it more scientifically coherent. But I'm pretty sure the point I meant to make can be effectively made without doing that. If people who haven't already made up their minds about whether I even have a point worth further considering, I'll appreciate their help trying to make that point a little more clear as needed, just as I appreciate your help. 

>Social scientists may have protocols for testing hypothesis regarding motivations but I'm way skeptical of any such methodologies. In fact, I don't even regard the social 'sciences' as being experimental sciences at all.

I do think it's unfortunate that I stand to be construed as trying to see into people's minds. That continues to be a major problem with psychology and social sciences.

The reason I'm offering the malefactor model here is that it has proven to have a lot of predictive value over the last 6 months and more. As a scientific person, I would really like to see some competing models. I'm really not all that interested in the question of whether there's some kind of counterproductive effort with LL, except in terms of how taking that as a predictive assumption produces good predictive results. 

Psychology and the social sciences unfortunately continue to use behavior terminology that is infected with nonbehavioral implications. 

It's difficult to distinguish, for example, between the claim that someone displays no empathy and the statement that someone feels no empathy. The fact that the word "empathy" can even be used in both cases merely underscores the ease with which one can become confused about the point being made. 

The malefactor I have hypothesized to be operating inside LL would not necessarily have to feel anything in particular; he would only need to produce a pattern of behavior predictably consistent with an intention to do harm to the interests of merchants; that is: a trend of harmfulness to significantly exceeding that at the chance level.

One example of this could be the apparent asymmetry in transaction errors. Chance error should produce statistical symmetry. Instead we have an error such that money repeatedly gets lost or destroyed, but never seems to be produced out of nothing by means of a transaction error. At the least sophisticated analytic level, these transaction errors are a benefit to LL or someone at LL, because, even if the money isn't going into a specific Linden's pocket, it's money that never has to be paid out later. At the same level of analytic sophistication, these transactions are malevolent to merchants. Even if the item is also not delivered, the transactions represent a disruption to relationships with customers (and they're bad for customers, too).

A fair question is whether or not it would really be in LL's interest to keep taking this money. While our intuition is that it is probably not, we can't be certain without proper analysis of the real data over time. So whether anyone at LL intends to allow it to continue is unclear. They could be very confused, they could be helpless to stop it, or they could think it's a great idea.

Where the malefactor model comes in an example like this is in terms of telling us whether we should expect the behavior to continue, and whether we should expect to see anything being done about it.

If there's no intention behind stopping it or allowing it to continue, we should expect it to continue, but we should also expect nothing to be said about it.

If there's an intention to stop it, we should see an acknowledgement of the problem and not a denial.

If there's an intention to allow it to continue, we should see a denial.

So far, what we're seeing is least inconsistent with which predictive model?

If we picked the right model a long time ago, we would at least have had the right prediction, and if there were some way to adjust our own behavior to make use of the prediction, we could have done that. 

And, in my case, I actually have made decisions that were importantly informed by this prediction, made by using this predictive model, and they have been good decisions.

Can I see into Lindens' minds?

Absolutely NO.

Does pretending, is some way that that's sort of possible ever produce any useful results?

Absolutely YES.

> Actually, I'd attribute more malfeasance to the "moderation" of these fora than I would to any perceived attempt to undermine the MP for whatever sinister reasons. Seems to me like there's an active effort being made to run off the most interesting & flamboyant posters.

Yes, and no.

Only very specific topics really tend to elicit moderation efforts.

This is another thing I have tested with some appreciable degree of rigor. 

The moderation service is not being exploited by just any whiny, cranky person. It is being exploited by a person or persons who want specific things not to be said, ever, in any context or with any kind of spin. 

What moderation efforts tend to have in common is that they cause to be removed from discussion some thing which a Linden has done which a Linden very clearly should not do.