The Visual Editor is a huge failure

[dogbiscuit]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

Russavia heckles and jayvdb tries, in vain, to get some straight answers while Maggie Dennis and Elitre babble inanely.

TL;DR - VE is used by very few people even two years after deployment.

Some interesting pages fall out of the discussion:
Charts, charts and more charts about VE
https://meta.wikimedia.org/wiki/Research:VisualEditor

A list of victims
https://www.mediawiki.org/wiki/VisualEditor/Rollouts

Note that the Dutch, English, German and Spanish wikis do not have VE enabled "by default".
What percentage of total edits and total article count are those four against the rest of the pile of little wikis?

Be interesting to see an update on the results of the impact on new editors. At least the analysis was honest and determined that VE actually did make it worse for new editors compared with the old editor.

https://meta.wikimedia.org/wiki/Researc ... rs/Results

The date of the experiment was about a year ago, and it would be interesting to see if the results of this could be correlated to any change in attitude from "force the bastard in and fuck 'em!" to "Shhh! If we keep our heads down and don't argue too much people might not notice that it is a steaming pile of do-do's"

[Randy from Boise]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

Russavia heckles and jayvdb tries, in vain, to get some straight answers while Maggie Dennis and Elitre babble inanely.

TL;DR - VE is used by very few people even two years after deployment.

Some interesting pages fall out of the discussion:
Charts, charts and more charts about VE
https://meta.wikimedia.org/wiki/Research:VisualEditor

A list of victims
https://www.mediawiki.org/wiki/VisualEditor/Rollouts

Note that the Dutch, English, German and Spanish wikis do not have VE enabled "by default".
What percentage of total edits and total article count are those four against the rest of the pile of little wikis?

Be interesting to see an update on the results of the impact on new editors. At least the analysis was honest and determined that VE actually did make it worse for new editors compared with the old editor.

https://meta.wikimedia.org/wiki/Researc ... rs/Results

The date of the experiment was about a year ago, and it would be interesting to see if the results of this could be correlated to any change in attitude from "force the bastard in and fuck 'em!" to "Shhh! If we keep our heads down and don't argue too much people might not notice that it is a steaming pile of do-do's"

Bottom line: The buggy early version of VE was a detriment to productive editing, not a boon.

WMF should be forced to demonstrate that VE is significantly better than wikitext before they try making it the default for new users again.

RfB

[Vigilant]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

Russavia heckles and jayvdb tries, in vain, to get some straight answers while Maggie Dennis and Elitre babble inanely.

TL;DR - VE is used by very few people even two years after deployment.

Some interesting pages fall out of the discussion:
Charts, charts and more charts about VE
https://meta.wikimedia.org/wiki/Research:VisualEditor

A list of victims
https://www.mediawiki.org/wiki/VisualEditor/Rollouts

Note that the Dutch, English, German and Spanish wikis do not have VE enabled "by default".
What percentage of total edits and total article count are those four against the rest of the pile of little wikis?

Be interesting to see an update on the results of the impact on new editors. At least the analysis was honest and determined that VE actually did make it worse for new editors compared with the old editor.

https://meta.wikimedia.org/wiki/Researc ... rs/Results

The date of the experiment was about a year ago, and it would be interesting to see if the results of this could be correlated to any change in attitude from "force the bastard in and fuck 'em!" to "Shhh! If we keep our heads down and don't argue too much people might not notice that it is a steaming pile of do-do's"

Bottom line: The buggy early version of VE was a detriment to productive editing, not a boon.

WMF should be forced to demonstrate that VE is significantly better than wikitext before they try making it the default for new users again.

RfB

They haven't though.
They exported their disease to third world wikis.

Sort of like American tobacco companies after cigarette lawsuits in the USA.

[Kiefer.Wolfowitz]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

Russavia heckles and jayvdb tries, in vain, to get some straight answers while Maggie Dennis and Elitre babble inanely.

TL;DR - VE is used by very few people even two years after deployment.

Some interesting pages fall out of the discussion:
Charts, charts and more charts about VE
https://meta.wikimedia.org/wiki/Research:VisualEditor

A list of victims
https://www.mediawiki.org/wiki/VisualEditor/Rollouts

Note that the Dutch, English, German and Spanish wikis do not have VE enabled "by default".
What percentage of total edits and total article count are those four against the rest of the pile of little wikis?

[It would be] interesting to see an update on the results of the impact on new editors. At least the analysis was honest and determined that VE actually did make it worse for new editors compared with the old editor.

https://meta.wikimedia.org/wiki/Researc ... rs/Results

The date of the experiment was about a year ago, and it would be interesting to see if the results of this could be correlated to any change in attitude from "force the bastard in and fuck 'em!" to "Shhh! If we keep our heads down and don't argue too much people might not notice that it is a steaming pile of do-do's"

Alas, this study seems to have been an observational study rather than a randomized experiment.

Research:VisualEditor's effect on newly registered editors/Results

Methods:
To test the effects of VisualEditor, we performed a controlled test on the English Wikipedia. During an 86-hour assignment period, newly registered user accounts were randomly bucketed (round robin) into two experimental conditions.

• Control: New users who received an odd user ID on registration were given an experience that matches the way that Wikipedia worked before VisualEditor was deployed, i.e. the default wikitext editor.
  Test: New users with an even user ID had VisualEditor enabled and could edit in VisualEditor (via the Edit link) or in wikitext (via the Edit source link). Section edit links for users in this condition opened the VisualEditor screen (see browser support).

Their "test" seems to have been an observational study. The association of editors to treatments may have been haphazard, that is, outside the control of the experimenters and plausibly without obvious bias.

It seems that the editors were not randomly assigned to treatments ("experimental conditions", that is, either VE or Wikitext) and the assignment was not controlled. If the WMF researchers had wished to perform a randomized experiment, before the week of the study, then they could have randomly generated assignments, which would have been stored in a list. As each new editor arrived, their assignment to either VE or Wikitext could have been read from the list.

Observational studies are much weaker than controlled randomized experiments.

Update: I tried to clarify the description of the study,

(→‎Methods: Clarify that editors haphazardly received id-numbers, which determined their exposure to VE or Wikitext. There was no random assignment or experimental control. These are common mistakes of non-statisticians trying to do experiments.)


Old: "we performed a controlled test on the English Wikipedia. During an 86-hour assignment period, newly registered user accounts were randomly bucketed (round robin) into two experimental conditions.

KW: "we conducted an observational study on the English Wikipedia. During an 86-hour study period, newly registered user accounts were haphazardly associated with two exposures (analogous to the treatments of controlled experiments).
...
In other words, the haphazard assignment of id-numbers to newly arriving editors determined exposure (as control or test). The researchers did not control the assignment of control- or test-treatments to editors; in particular, the researchers did not randomly assign treatments to editors. The study was an observational study."

and I left a note on the talk page.

Observational study with haphazard exposure to VE or Wikitext

The researchers should have generated a list of random assignments, which would have been sequentially assigned to new editors along with their user id.

Instead, the parity of the user id (either odd or even) determined the exposure to VE or Wikitext. While this association was haphazard and without obvious bias (imho), it was not under the control of the researchers and was not randomized.

You should consult with a competent statistician with experience in experiments. Kiefer.Wolfowitz 17:49, 21 June 2014 (UTC)

[Vigilant]

I wonder what the dollars spent per byte of editing ($/B) ratio is for the VisualEdsel...

In most tools, you'd expect the amortization of massive edit counts over a dwindling development and sustaining load to move that cost towards zero.

In VisualEdsel's case, I expect the cost is increasing as they add headcount, development and support for lots of wikis with very, very low edit counts.

[The Adversary]

After one year, I tried to use the VE again last week. I was just trying to do something very simple; (adding a page-number to a reference). I spent 15 minutes, going "$$%&&/%$$##!"!$&
Never managed to do a thing.


(And when I have looked at the 500 recent changes on en.wp recently, VE typically has been used in 0 to 4 edits)

(...and apparently we are violating "The friendly space" policy if we ask why the whole VE-team is not fired...)

[Hex]

[15:15:22] <James_F> Yeah, we're relatively aggressive in fixing bugs we find, so often people won't notice the issue before it's already solved.

[Vigilant]

[15:15:22] <James_F> Yeah, we're relatively aggressive in fixing bugs we find, so often people won't notice the issue before it's already solved.

Yeah.....

[mac]

[15:15:22] <James_F> Yeah, we're relatively aggressive in fixing bugs we find, so often people won't notice the issue before it's already solved.

Yeah.....

I thought those were "surfaced" by The Community™, although to be fair, the community liaisons were rather aggressive.

[Anthonyhcole]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

I've pulled out the John Vandenberg-James Forrester discussion about user adoption of the visual editor:

it has been a year since launch ? what is the current percentage of edits using VE ?

Hey. Yes, it's just under a year since the initial "on by default" switch for a big set of wikis. It's been just over two years since VisualEditor went live with original community testing (initially on MediaWiki.org). The usage of VisualEditor is mixed between different wikis, and different user groups, but to give a brief summary:
On wikis where VisualEditor is on by default, anonymous users use VisualEditor for ~ a third of their content edits. Users of registered accounts created after the switch to default use it for ~ a fifth. Users of registered accounts created before the switch to default ("old hands" or whatever) use it for ~ a twentieth. So, as you can see, adoption isn't as high as we'd want and we're looking to make VisualEditor better so that more people choose to use it over wikitext. (The alternative of course would be to make wikitext worse, but that wouldn't be very ethical. ;-))

do you have any targets for adoption?

However, slightly more anonymous users use VisualEditor in any given day than those who use wikitext. So to speculate, the old-hand IP editors (who we know exist but can't label in our data) look like they're more likely to use wikitext, and more likely to be productive in the number of edits. But we don't know for sure.

In terms of targets (to answer question three before Elitre can do question two?), ish. "Higher", certainly. I think we can probably convert IPs over to a 2:1 ratio, at least.

your responses are a bit confusing. first, do you have an overall percentage of edits on all wikis? secondly, "anonymous users use VisualEditor for ~ a third of their content edits" vs "slightly more anonymous users use VisualEditor in any given day than those who use wikitext" dont line up - could you explain how these two statements are both accurate.

On all wikis? I could ask Analytics to crunch that number, but I don't have it to hand. On wikis where VisualEditor is active, I believe that VE is ~ 6.5% of all edits (but that counts talk page edits, bot edits, etc.) so it's not really a valuable number to talk about. Post-Flow maybe we can have a better idea, but…

I would appreciate it if you could provide that statistic in future office hours. I will ask it again ;-)

As I said, that number doesn't exist. We could ask Analytics to calculate it instead of other work they're doing, if you think it's really critical. I don't.

with or without the caveat "on wikis where VisualEditor is active" which is IMO a cop-out as it is bugs preventing further rollout, so % of edits being a reasonable measure of effectiveness of this project, wikis it cant be enabled on are non-effectiveness

For example, no Wikidata edits are made with VisualEditor (and can't be); skewing the data with millions of spurious entries is misleading at best.

ok, then the caveat should be: in all namespaces where the content model is wikitext

Filtering out bot edits, bot-like edits, talk-page edits, LiquidThread edits, Wikidata edits and hundreds of other non-VE-possible edits is probably not a trivial matter to ask Analytics to spend time on instead of more important questions like "what is the effect of new data centre our reading speeds for uses in Asia" or similar.

James doesn't think tracking and reporting the effectiveness of his project is important. No target. No tracking. No accountability. Lucky guy.

Does anyone know which Wikipedias are using this as the default editor for newbies?

[EricBarbour]

James doesn't think tracking and reporting the effectiveness of his project is important. No target. No tracking. No accountability. Lucky guy.

Meh, he simply needs to be slapped in the mouth more often....

Does anyone know which Wikipedias are using this as the default editor for newbies?

Far as I can tell, no one knows. No statistics are being kept, in typical Wikimedia fashion.

[Vigilant]

http://en.wikipedia.org/w/index.php?tit ... ematics_II

Another nail in the coffin.

Here's a shout out to James Forrester and Erik Mo:eller.

[Vigilant]

[15:15:22] <James_F> Yeah, we're relatively aggressive in fixing bugs we find, so often people won't notice the issue before it's already solved.

Yeah.....

I thought those were "surfaced" by The Community™, although to be fair, the community liaisons were rather aggressive.

I can imagine the WMF community "liaisons" issuing quotas to the VE users like Colombus to the Taino people...

[Vigilant]

One of the rules of our software development is that if your answer is "the input is wrong", you're starting from the wrong place. :-) Asking for the English Wikipedia community (and hundreds of others) to adjust millions of pages to make our software work better because we couldn't think of a proper way to fix the problem doesn't seem like the best solution (not that I have an actionable alternative right now). :-( Jdforrester (WMF) (talk) 01:08, 25 June 2014 (UTC)

You sad, fat little man.
Asking for the English Wikipedia community (and hundreds of others) to adjust millions of pages to make our software work better to a broken new editor because we couldn't think of a proper way to fix the problem build one doesn't seem like the best solution...

[Kiefer.Wolfowitz]

I tried to clarify the description of the study,

(→‎Methods: Clarify that editors haphazardly received id-numbers, which determined their exposure to VE or Wikitext. There was no random assignment or experimental control. These are common mistakes of non-statisticians trying to do experiments.)

Old: "we performed a controlled test on the English Wikipedia. During an 86-hour assignment period, newly registered user accounts were randomly bucketed (round robin) into two experimental conditions.
KW: "we conducted an observational study on the English Wikipedia. During an 86-hour study period, newly registered user accounts were haphazardly associated with two exposures (analogous to the treatments of controlled experiments).
...
In other words, the haphazard assignment of id-numbers to newly arriving editors determined exposure (as control or test). The researchers did not control the assignment of control- or test-treatments to editors; in particular, the researchers did not randomly assign treatments to editors. The study was an observational study."

and I left a note on the talk page.

Observational study with haphazard exposure to VE or Wikitext

The researchers should have generated a list of random assignments, which would have been sequentially assigned to new editors along with their user id.

Instead, the parity of the user id (either odd or even) determined the exposure to VE or Wikitext. While this association was haphazard and without obvious bias (imho), it was not under the control of the researchers and was not randomized.

You should consult with a competent statistician with experience in experiments. Kiefer.Wolfowitz 17:49, 21 June 2014 (UTC)

WMF employee Aaron Halfaker Halfak (WMF) (T-C-L) labeled my edits "non-constructive" and reverted them.

Reverted your edits. As you admit, there's no apparent bias to the bucketing strategy and therefor the statistical validity of the study. Your changes to the copy were less than helpful. If you would like to bring forward substantial concerns about the potential for bias in the described study, I welcome such a discussion. --Halfak (WMF) (talk) 19:15, 23 June 2014 (UTC)

I replied on the talk page:

@Halfak (WMF):
To clarify:

• 1. There was no randomized assignment of users to treatments. The study relied on the haphazard arrival of users and their labeling to determine exposure.
  2. You assert that an observational study using haphazard exposures have "statistical validity".
  3. You are putting words in my mouth and, I'd like to think, misreading what I wrote. That I cannot name a bias in the haphazard exposure does not mean that none exist. Guarding against unforeseen biases is one of the reasons competent experimenters randomize when possible.

Please revert your reversion.
Kiefer.Wolfowitz 22:05, 25 June 2014 (UTC)

A proper response would have thanked me for clarifying the study and admitted "yes, we should have randomized the assignment of editors to VE or Wikitext, which could have been done easily as you outlined".

Aaron's en:WP user page lists his "papers" (weak conference proceedings), including this one establishing Halfak's claim to expertise on reverting new editors, such as myself on Meta:
Halfaker, A., Kittur, A., & Riedl, J. (2011). Don't Bite the Newbies: How Reverts Affect the Quantity and Quality of Wikipedia Work, WikiSym’11. ACM. (pdf)

[Vigilant]

Some of his musings
http://en.wikipedia.org/w/index.php?tit ... iewed_work

[EricBarbour]

WMF employee Aaron Halfaker Halfak (WMF) (T-C-L) labeled my edits "non-constructive" and reverted them.

Reverted your edits. As you admit, there's no apparent bias to the bucketing strategy and therefor the statistical validity of the study. Your changes to the copy were less than helpful. If you would like to bring forward substantial concerns about the potential for bias in the described study, I welcome such a discussion. --Halfak (WMF) (talk) 19:15, 23 June 2014 (UTC)

I replied on the talk page:

@Halfak (WMF):
To clarify:

• 1. There was no randomized assignment of users to treatments. The study relied on the haphazard arrival of users and their labeling to determine exposure.
  2. You assert that an observational study using haphazard exposures have "statistical validity".
  3. You are putting words in my mouth and, I'd like to think, misreading what I wrote. That I cannot name a bias in the haphazard exposure does not mean that none exist. Guarding against unforeseen biases is one of the reasons competent experimenters randomize when possible.

Please revert your reversion.
Kiefer.Wolfowitz 22:05, 25 June 2014 (UTC)

A proper response would have thanked me for clarifying the study and admitted "yes, we should have randomized the assignment of editors to VE or Wikitext, which could have been done easily as you outlined".

There are some things you could ask him:

1) How many of those new accounts were sockpuppets generated with unregistered bots, and not merely "autocreated for users on other wikis"? Do you even know which ones are these mechanically-generated socks?
2) Did you know that 60% of all registered Wikipedia user accounts never do ANY editing, at all? And that most of those accounts appear to be autogenerated sockpuppets, held in reserve by persons unknown for unknown reasons?
3) Did you allow for these non-editing sockpuppets in your study?
4) How many of the new users were only committing vandalism, and how were they handled statistically?

[Poetlister]

You sad, fat little man.

That's irrelevant; do you think he'd be any better if he lost weight?

[Vigilant]

You sad, fat little man.

That's irrelevant; do you think he'd be any better if he lost weight?

Lectures from you?

How droll.

[Kiefer.Wolfowitz]

WMF employee Aaron Halfaker Halfak (WMF) (T-C-L) labeled my edits "non-constructive" and reverted them.

Reverted your edits. As you admit, there's no apparent bias to the bucketing strategy and therefor the statistical validity of the study. Your changes to the copy were less than helpful. If you would like to bring forward substantial concerns about the potential for bias in the described study, I welcome such a discussion. --Halfak (WMF) (talk) 19:15, 23 June 2014 (UTC)

I replied on the talk page:

@Halfak (WMF):
To clarify:

• 1. There was no randomized assignment of users to treatments. The study relied on the haphazard arrival of users and their labeling to determine exposure.
  2. You assert that an observational study using haphazard exposures have "statistical validity".
  3. You are putting words in my mouth and, I'd like to think, misreading what I wrote. That I cannot name a bias in the haphazard exposure does not mean that none exist. Guarding against unforeseen biases is one of the reasons competent experimenters randomize when possible.

Please revert your reversion.
Kiefer.Wolfowitz 22:05, 25 June 2014 (UTC)

A proper response would have thanked me for clarifying the study and admitted "yes, we should have randomized the assignment of editors to VE or Wikitext, which could have been done easily as you outlined".

There are some things you could ask him:

1) How many of those new accounts were sockpuppets generated with unregistered bots, and not merely "autocreated for users on other wikis"? Do you even know which ones are these mechanically-generated socks?
2) Did you know that 60% of all registered Wikipedia user accounts never do ANY editing, at all? And that most of those accounts appear to be autogenerated sockpuppets, held in reserve by persons unknown for unknown reasons?
3) Did you allow for these non-editing sockpuppets in your study?
4) How many of the new users were only committing vandalism, and how were they handled statistically?

He must walk before he can run.
The discussion continues. He is claiming that his study is "unbiased" even without controlled assignment or randomization.

Hey Kiefer.Wolfowitz.
Assignment in this test was not haphazard; as the text suggests, a round-robin bucketing strategy was used. This provides for unbiased and even assignment to the experimental conditions. I leave it up to you to explain how the this assignment strategy might bias the results. At the very least, I would appreciate if you could cite something suggesting that round-robin assignment is problematic.
--Halfak (WMF) (talk) 00:33, 26 June 2014 (UTC)

Halfak (WMF)
You did not use randomization. Would you please explain any sense that your "unbiased" have?
In randomized samples and experiments, the objective randomization specified in the protocol induces an objective probability distribution on the outcomes, of which one is observed in a particular study. With respect to this distribution, unbiased estimates of population parameters and of treatment effects have been studied.
In particular, my and your opinion of the subjective likelihood of a bias due to your failure to randomize is irrelevant. There is no objective basis for your "unbiased" claim. Textbooks on randomized experiments typically discuss hidden biases that became apparent upon autopsy of a failed experiment, in cases where the experiment was actually of interest to other researchers, rather than Potemkin science for write-only conferences. One example I recall was an animal experiment in which there was a shelf effect on litters (each kept on cages), which was only discovered after a failed experiment. I suppose the International Cancer Society's 3rd volume on long-term animal experiments has a discussion of the importance of randomization, even within a cage. Assigning treatments to mice based on the haphazard order an assistant pulls them out of cages is not recommended.
In observational studies, a probability model may also conjectured. A parametric model may be taken seriously if a model has been previously validated, although such examples are rare. If somebody pretends that the data be a random sample from a distribution (with a subjective probability model), then some unbiased estimates may also be available. (However, such subjective methods are weaker, as students are warned in basic courses.) So, are you pretending that the editors were a random sample from a distribution? Parametric?
You might look at David A. Freedman et alia's Statistics for warnings about non-randomized studies. Kempthorne and Hinkleann's experimental design book is thorough. Speaking with authority, John Tukey advised experiments "to randomize unless you are a damned fool".
Exposing editors to a predictor (wikitext or VE) based on their id being even or odd is trivial. There are non-trivial round-robin designs discussed e.g. in Ian Anderson's book on combinatorial designs.
Kiefer.Wolfowitz (talk) 11:53, 26 June 2014 (UTC)

He does not understand the distinction between random and haphazard.

[dogbiscuit]

He does not understand the distinction between random and haphazard.

To be fair, the exercise is so pointless that the sampling is probably the least of the problems. The total inability to understand the population being analysed is a better starting point.

[Kiefer.Wolfowitz]

He does not understand the distinction between random and haphazard.

To be fair, the exercise is so pointless that the sampling is probably the least of the problems. The total inability to understand the population being analysed is a better starting point.

Good point, but your terminology of "sampling" is problematic. Random sampling is used in surveying populations. Randomized assignment of experimental units to treatments is used to study treatment effects, in which case it is better to avoid the term "sampling".

A small experiment is often worth doing, to provide insight into causal mechanisms, even if the experimental units (subjects) are not a random sample from the general population. Once a treatment effect has been demonstrated for a reasonable experiment, one can begin to worry about representativeness.

In sometimes harmful econometrics and statistics, a conventional analysis assumes that there is a constant treatment effect, which is the same for all experimental units (subjects). That strong assumption, if plausible, would reduce the wish for a representative or random sample of experimental units from the population of interest, since the effect be the same for everybody!

More generally, a model could have an average treatment-effect, in which case the individual treatment-effects are random variables; having zero variance implies that the random variables be constant (almost surely), as assumed in the conventional analysis.

[thekohser]

You sad, fat little man.

That's irrelevant; do you think he'd be any better if he lost weight?

Lectures from you?

How droll.

Isn't "little" a bit of a heightist thing to say? And don't get me started on "man" -- why wasn't "person" sufficient? Obviously sexist.

[Johnny Au]

Hidden comments will be viewable and editable, as the VE engineers are working on them.

[Vigilant]

Here's a world beater...

True section editing—in which you read everything, but then discard all the things that are irrelevant (e.g., images that don't extend into this section)—solves this problem. However, it is unavoidably slower. Here's the current system, which most advocates for section editing have already deemed to be too slow:

Send everything to your computer (two seconds).
Process everything locally (five seconds).
Make your change (however long you want).
Send everything back (three seconds).

Here's what true section editing would do:

Send everything to your computer (two seconds).
Process everything locally (five seconds).
Throw away three-quarters of what you just processed (two seconds).
Make your change (however long you want).
Send everything back (two seconds).

There's a net loss to speed here.
Also, before anyone asks again, section editing hasn't prevented edit conflicts for about eight years now. Whatamidoing (WMF) (talk) 00:54, 1 July 2014 (UTC)

Firstly, it takes you TWO FUCKING SECONDS to throw stuff away?!?!

These numbers are wildly out of line with what I would expect for a project like this.
The average article on wikipedia is what? 30K? Assuming a 1 meg connection should be on the order of 100ms tops.
Do they reload the VisualEdsel on every edit?! No caching on the user's machine?!
Why not just check on every edit instance if there's an update for VisualCowpat available and download it then?

Why in the hell are they processing everything on the user's machine?
Is the local representation that much larger than the stored version?!
FIVE SECONDS to process a 30K article?!?!?!
Whiskey. Tango. Foxtrot?!
SIX characters per millisecond?!
I'm flabbergasted.
Why not preprocess the article on the server, caching it in a serializable format than can be economically squirted to the client?

Why does sending everything back take TWO WHOLE SECONDS?
Why is this a symmetric amount of time to loading the entire article?!
Surely they just send back a delta of the edit?

The more that WMF engineering speaks, the less show they know.

[Vigilant]

AHAHHAHAHAHAHAHHAHAHHAHAHHAHAHHAHAHHAHAHA

New England

Here's what happened when I tried to edit New England using VE. I clicked Editbeta and four blank lines appeared before the lead paragraph. I positioned my cursor at the beginning of the lead paragraph (to the left of the "N" in "New England") and pressed Backspace a few times to delete the blank lines:

the first backspace appeared to do nothing
the second backspace deleted one of the four blank lines
the third backspace deleted the lead paragraph (!)
subsequent backspaces did nothing.

Windows 7/Firefox 30.0, reproducible each time I tried. 28bytes (talk) 00:32, 1 July 2014 (UTC)

Yes, I get the same (also W7 FF30), so not some 28bytes-specific problem. Note that the second backspace, which only looked as if it removed a blank line, in reality removed the File:Chestnut Street Salem, which is also unwanted (but already mentioned) behaviour... The third backspace removed some more images and moved the first paragraph inside the infobox, believe it or not, and all subsequent backspaces were edits inside the infobox! Fram (talk) 07:33, 1 July 2014 (UTC)

Totally something you COULD NEVER CATCH IN REGRESSION TESTING!!!!

Dumbasses.
Complete dumbasses.

Two years in with copious feedback and a monstrously large team, for the relative difficulty of the problem they were attempting to solve, and they can't catch even the most basic problems before releasing this, still alpha quality, ... product.

The WMF really, really, really needs to halt development on its most troubled projects, VE, FLow, Media Viewer, and reassess whether continuing to POUR money into these abysses without any end in sight is the best use of charitable donations.

[lilburne]

It doesn't matter what the WMF do, page editing software WILL NOT significantly increase the number of people editing pages.

A couple of years ago I installed phpBB software on a webhost for a group of people that wanted a more permanent record of stuff than that available on FB. After an initial flurry of interest it died. Not because they weren't interested but because the mechanics of editing a post in forum software is too difficult. Simply put they could not work out how to post a link, how to embed an image or video, and the formatting controls were just screwy. Perhaps it was the drugs I dunno, but on FB they C&P the link and it works they don't have to care about whether it is a video, image, or webpage, a nice little preview is provided. They just type shit and press enter. They don't have to care about bold, underline, or any of that other stuff. This is a major reason why FB is popular, you don't have to know, you don't have to care.

I'm 100% certain that unless the MOS is automatically enforced in the software, it alone will deter any significant increase in the number of editors.

[Vigilant]

Oh goodie, here we go again!
https://www.mediawiki.org/w/index.php?t ... 43#Editing

Editing

Goal-setting process owner: James Forrester

The Editing Team is working to make VisualEditor a great editor for new and experienced editors alike, focusing on improving the performance and usability whilst adding some more features to make VisualEditor more helpful, intuitive and practical for use for every content edit, alongside maintaining, improving, and extending the existing editor software. Below are a set of goals that we hope to achieve over the next financial year (July 2014 – June 2015), with a balance of optimistic and pessimistic assumptions about speed of delivery; not all goals may be achieved in the time period indicated, and some may be updated over time.

On-going work happening every quarter:

Stability and bug fixing, prioritising any bugs that cause wikitext corruption or any other form of disruption for our wikis’ communities (including in the wikitext editor);
Performance improvements for users, tracking load & save times, execution speed;
UX improvements tracked regularly and reported in a quarterly public user testing narrative;
Ensuring the success of VisualEditor on mobile, with a target of feature equivalence on tablet and at least some features on phone, with responsibility for both VisualEditor and wikitext editing pipelines on tablets and phones as well as desktop transitioning from Mobile to the team; and
Collaborating with volunteers and other Engineering teams like Parsoid, Services, Platform, & Core on related efforts like skin improvements, front-end performance, and other areas.

Quarter Goals
Jul–Sep 2014

Deployment: Engaging with English Wikipedia to discover pain points as part of agreeing the criteria for a gradual ramp-up of VisualEditor availability and usage to default, expected to happen in Q2, subject to community discussions (after this point, ongoing support)
Core: Internet Explorer 9/10+ browser support
Feature: Auto-filling citations from ISBN, DOI or URL
Feature: Editing templates’ parameters as rich content, not wikitext, with helper tools for some types like image (searching Commons), link (searching wikis), date (date selector extended from Wikidata’s), and possibly others
Metrics:
Load performance: No significant changes expected – 3s 50%ile, 5s 75%ile, 25s 99%ile (same as baseline)
Save performance: Major improvement is compressing save data before submitted for save – 3s 50%ile, 6s 75%ile, 15s 99%ile (from 4s 50%ile, 8s 75%ile, 15s 99%ile baseline)
Per-edit adoption: No significant changes expected – 35% IPs, 20% post-default users, 4% pre-default users (same as baseline)

Oct–Dec 2014

Deployment: Engaging with German, Dutch and Spanish Wikipedias in a similar fashion to English Wikipedia, agreeing the ramp-up to default to happen from around Q3 onwards, subject to community discussion
Core: Initial improved support for IMEs, for key expanded if not all language groups
Feature: Table editing – inserting new and deleting existing rows and columns

Jan–Mar 2015

Deployment: Engaging with non-Wikipedias to consider issues for them like which key extensions and gadgets will need VisualEditor support
Feature: Uploading media – uploading an image/video/etc. to Commons mid-edit via a button in the toolbar, inserting into the edited page on completion
Feature: Wikidata invocation editing as a template parameter type

Apr–Jun 2015

Core: Language variant editing support for Chinese and other languages
Core: Release a stand-alone VisualEditor for third party embed-ability(HAHHAHAHHAHAH)
Feature: Auto-save local-only drafts if a browser crashes/disconnects mid-edit
Feature: Additional table editing – common "advanced" tools like sortable columns & table header

Interdependencies:

Auto-filling citations depends on the Outreach Program for Women work to do this led by the Services team (and partially mentored by the Editing team).
Uploading media functionality relies on work planned to be done in the Multimedia team.
Language variants support relies entirely on work planned to be done in the Parsoid team.
Continued key on-going dependencies on the Parsoid, Services, Platform Core MediaWiki and Mobile teams, and collaboration with the Core Features and Growth teams for their VisualEditor-related goals

Guys.
Stop.
They don't want your shit editor.

Are you going to go the SOPA route and pose the question in so many guises over so much time that you think you'll trick them into agreeing once and then it'll be locked in forever?

[Kiefer.Wolfowitz]

Vigilant quoted:

Auto-filling citations depends on the Outreach Program for Women work to do this led by the Services team (and partially mentored by the Editing team).

[EricBarbour]

Oh goodie, here we go again!
https://www.mediawiki.org/w/index.php?t ... 43#Editing

Are you going to go the SOPA route and pose the question in so many guises over so much time that you think you'll trick them into agreeing once and then it'll be locked in forever?

As long as Forrester is on the WMF staff, I expect they'll keep trying. Appears to be "his baby" now.

[Randy from Boise]

James Forrester takes a turn in the barrel.
https://meta.wikimedia.org/wiki/IRC_off ... 2014-06-19

I've pulled out the John Vandenberg-James Forrester discussion about user adoption of the visual editor:

it has been a year since launch ? what is the current percentage of edits using VE ?

Hey. Yes, it's just under a year since the initial "on by default" switch for a big set of wikis. It's been just over two years since VisualEditor went live with original community testing (initially on MediaWiki.org). The usage of VisualEditor is mixed between different wikis, and different user groups, but to give a brief summary:
On wikis where VisualEditor is on by default, anonymous users use VisualEditor for ~ a third of their content edits. Users of registered accounts created after the switch to default use it for ~ a fifth. Users of registered accounts created before the switch to default ("old hands" or whatever) use it for ~ a twentieth. So, as you can see, adoption isn't as high as we'd want and we're looking to make VisualEditor better so that more people choose to use it over wikitext. (The alternative of course would be to make wikitext worse, but that wouldn't be very ethical. ;-))

do you have any targets for adoption?

However, slightly more anonymous users use VisualEditor in any given day than those who use wikitext. So to speculate, the old-hand IP editors (who we know exist but can't label in our data) look like they're more likely to use wikitext, and more likely to be productive in the number of edits. But we don't know for sure.

In terms of targets (to answer question three before Elitre can do question two?), ish. "Higher", certainly. I think we can probably convert IPs over to a 2:1 ratio, at least.

your responses are a bit confusing. first, do you have an overall percentage of edits on all wikis? secondly, "anonymous users use VisualEditor for ~ a third of their content edits" vs "slightly more anonymous users use VisualEditor in any given day than those who use wikitext" dont line up - could you explain how these two statements are both accurate.

On all wikis? I could ask Analytics to crunch that number, but I don't have it to hand. On wikis where VisualEditor is active, I believe that VE is ~ 6.5% of all edits (but that counts talk page edits, bot edits, etc.) so it's not really a valuable number to talk about. Post-Flow maybe we can have a better idea, but…

I would appreciate it if you could provide that statistic in future office hours. I will ask it again ;-)

As I said, that number doesn't exist. We could ask Analytics to calculate it instead of other work they're doing, if you think it's really critical. I don't.

with or without the caveat "on wikis where VisualEditor is active" which is IMO a cop-out as it is bugs preventing further rollout, so % of edits being a reasonable measure of effectiveness of this project, wikis it cant be enabled on are non-effectiveness

For example, no Wikidata edits are made with VisualEditor (and can't be); skewing the data with millions of spurious entries is misleading at best.

ok, then the caveat should be: in all namespaces where the content model is wikitext

Filtering out bot edits, bot-like edits, talk-page edits, LiquidThread edits, Wikidata edits and hundreds of other non-VE-possible edits is probably not a trivial matter to ask Analytics to spend time on instead of more important questions like "what is the effect of new data centre our reading speeds for uses in Asia" or similar.

James doesn't think tracking and reporting the effectiveness of his project is important. No target. No tracking. No accountability. Lucky guy.

Does anyone know which Wikipedias are using this as the default editor for newbies?

I've been given a yellow card at Jimbotalk for talkin' truth about Flow. Somebody not named Tim should definitely bring this matter up there...

RfB

[HRIP7]

I've been given a yellow card at Jimbotalk for talkin' truth about Flow.

I see what you mean ...

MediaViewer is a proxy for the real fight, which as far as I am concerned is all about Flow — a real frankenstein monster created because the bureaucracy had a budget and needed something to do, with an absolutely gargantuan potential for disruption of the entire WP project. The fight needs to be fought (and lost) over the fairly benign MediaViewer — but a fight lost in such a way by a committed and aggressive ArbCom that Flow's damage is diverted from an English WP launch until it can be proven by practice elsewhere to be a substantial improvement. Obviously, there is inertia among those of us using the software leading to preference for old ways over new. Less obviously, there is a multimillion dollar careerist incentive for WMF Engineering to churn out something, anything new to justify their ever expanding budget. So, I beg to differ: YES, this is the hill where the fight needs to be made; because if the fight starts after Flow is already unilaterally imposed systemwide, the disruption will have already taken place and it will be too late. —Tim Davenport //// Carrite (talk) 17:24, 17 July 2014 (UTC) Last edit: Carrite (talk) 17:25, 17 July 2014 (UTC)

Carrite this is an extremely unfair and false representation of the situation that does nothing to bring light and love and healing and progress and does much to create the kind of tensions that make progress difficult. You've got the motivations exactly wrong. We badly need software improvements, including improvements to our rather ridiculous system of discussing things with each other by editing raw wikitext (instead of a philosophically wiki but more technically sophisticated approach that allows us to do the same things but in an easier and more intuitive way) and so we are investing engineering in that. It isn't like we are just showering money on the tech team for no reason and then they have to make up a reason to spend it. Saying that is just simply and purely a personal attack on good people who are doing good work. Stop doing that please.
Are there problems with disconnect between what editors want and what the developers are developing? Sometimes clearly yes. Sometimes that's because editors forget what readers want matters too. But sometimes it's just a dysfunctional disconnect and ALL SIDES have the capability through assuming good faith and entering into non-hostile dialog to work to change that. I encourage you to make your complaints in a positive and constructive way and point out better solutions. Insulting people and spreading FUD is just simply not ok.--Jimbo Wales (talk) 19:25, 17 July 2014 (UTC)

This is clearly not the venue for me to speak frankly on this matter. My apologies for attempting to do so. Carrite (talk) 20:16, 17 July 2014 (UTC)

[HRIP7]

You see, the way to bring light and love and healing™ is to tell critics to shut up, rather than try to understand their concerns.

[TungstenCarbide]

Carrite this is an extremely unfair and false representation of the situation that does nothing to bring light and love and healing and progress...

... and does much to create the kind of tensions that make progress difficult. You've got the motivations exactly wrong. We badly need software improvements, including improvements to our rather ridiculous system of discussing things with each other by editing raw wikitext (instead of a philosophically wiki but more technically sophisticated approach that allows us to do the same things but in an easier and more intuitive way) and so we are investing engineering in that. It isn't like we are just showering money on the tech team for no reason and then they have to make up a reason to spend it. Saying that is just simply and purely a personal attack on good people who are doing good work. Stop doing that please.
Are there problems with disconnect between what editors want and what the developers are developing? Sometimes clearly yes. Sometimes that's because editors forget what readers want matters too. But sometimes it's just a dysfunctional disconnect and ALL SIDES have the capability through assuming good faith and entering into non-hostile dialog to work to change that. I encourage you to make your complaints in a positive and constructive way and point out better solutions. Insulting people and spreading FUD is just simply not ok.--Jimbo Wales (talk) 19:25, 17 July 2014 (UTC)

Jimbo used to hate eric, going so far as to try to derail his board run. Now he's pumping eric's pet project that has languished and petered for years under different names.

[Randy from Boise]

You see, the way to bring light and love and healing™ is to tell critics to shut up, rather than try to understand their concerns.

This is a fucking car crash coming... Imposition of Flow will mean the automatic archiving of ALL discussion pages, so far as I am aware.

I have this great idea: if Flow is so terrific, prove it with a 60 day test at Jimbotalk. Let's see how good that gear works in practice...

t

Addenda: On a closely related note, it appears to me that W-- S------- has killed his Wiki trying to unilaterally impose Liquid Threads.

[Midsize Jake]

Auto-filling citations depends on the Outreach Program for Women work to do this led by the Services team (and partially mentored by the Editing team).

This bit of glib should probably read, "Auto-filling citations depends on work to be done under the auspices of the Outreach Program for Women, who have funded a summer intern to do this, who in turn will be led by the Services team (and partially mentored by the Editing team)."

That part of the proposal links to a subpage of User:Mvolz (i.e., Marielle Volz) on mediawiki.org, in which she outlines a Mediawiki extension that would produce properly formatted {cite|web} template code from a URL pasted directly from a browser tab. I'm sure it will be very handy, and indeed it's not a bad idea if you're into that sort of thing.

[Vigilant]

How long has this ... thing been in development?

Thanks for verifying that. It's not just the delete key; I've been experimenting and I can get bizarre behaviour in other ways. This rev, for example, seems to act up. I've repeatedly been able to get VE into a state where it starts randomly inserting characters without me touching the keyboard -- it appears to go into some kind of loop. This happens after making multiple edits to the same area mentioning above -- I used multiple del, backspace and Ctrl-I keys, as well as double clicking on a word to select it and typing over it. If I can get a keystroke sequence that will reliably reproduce it I will post it here. Mike Christie (talk - contribs - library) 22:22, 20 July 2014 (UTC)

Perfect.

And you bozos want to start trying to make this utter cowpat of a product the default editor on en.wp AGAIN? Soon?

What a bunch of putzes.

[Mason]

Some interesting numbers from Fram:

On the French Wikipedia, which is probably the largest wiki-version where VE is the default, the results are that after more than a year of VE deployment, less than 10% of the edits are made using the default editor (VE) and more than 90% are made with the cumbersome, old-fashioned, outdated, editor-frightening wikitexteditor.

source

[Poetlister]

Some interesting numbers from Fram:

On the French Wikipedia, which is probably the largest wiki-version where VE is the default, the results are that after more than a year of VE deployment, less than 10% of the edits are made using the default editor (VE) and more than 90% are made with the cumbersome, old-fashioned, outdated, editor-frightening wikitexteditor.

source

What a load of luddites!

[Vigilant]

Here we are, years and millions of dollars later, and the contents of the visual editor feedback page still look like this

1 Highlight disambiguation links and show all options on the disambiguation page for fixing them.
2 Autoupdate citation list
3 VE finally working in Internet explorer!
4 Table formatting modified by VE
5 References format modified by VE
6 Cut and paste of a table gives nowiki
7 Group refs question
8 nowiki around whitespace just before a colon

At some point, even the WMF is going to have to re-evaluate this project.

Can it ever be completed?
Can it ever be even moderately bug free?
Can it be maintained?
At what sustaining cost?

[The Garbage Scow]

I await the inevitable day when Wikipedia becomes a shameless, broken, third-rate knockoff of Twitter.

Jimmy Wales should read about Myspace sometime.

[EricBarbour]

I await the inevitable day when Wikipedia becomes a shameless, broken, third-rate knockoff of Twitter.

It became that back in 2005.

[Poetlister]

I await the inevitable day when Wikipedia becomes a shameless, broken, third-rate knockoff of Twitter.

It became that back in 2005.

Twitter doesn't allow 100k articles!

[Randy from Boise]

I await the inevitable day when Wikipedia becomes a shameless, broken, third-rate knockoff of Twitter.

It became that back in 2005.

Twitter doesn't allow 100k articles!

.........or 300k talk pages about them!

RfB

[Vigilant]

Three tries, three fails

Random article, first I got was Results of the Victorian state election, 2010 (Legislative Council). Open in VE, takes somewhat long, doesn't look like the original page. Click on the first box on the screen, click on the "Australian elections/Title row" box, template opens. Try to move the box to the right, so I can see the template (in the article) and the template edit box next to each other, so I know which parameter is feeding which "box" in the template. Can't move the template edit box. Give up. Try to find a method to edit the numbers per party inside the template. Can't find a method. Give up.

Second random article: Malamarismo. Try to edit the tracklist in VE. Parameters are given in alphabetical order, not in order of article. Useless. Give up.

Thrid try. Charles Landon Knight. Open the infobox (infobox congressman). "An infobox for office holders. It is generally better to use a more specific template like {{Infobox politician}}." Strange. Can I convert this to another infobox? Of course not. Do I need to? Of course not, this is a more specific template. If you would use infobox politician (like on Maurice Duverger), you would get the exact same message instructing you to use "Infobox politician" instead of, er, "Infobox politician". Why? Because template data aren't made with redirected sub-infobox-templates in mind, even though these are used very frequently. Give up.

Three tries, three fails. One case where I can't edit what I want, one where I can but it is easier and more user friendly in wikitext, and one where I get completely wrong advice which I can't easily follow in VE anyway. In all cases, I would have been better off editing in the wikitext editor (and in all cases, I needed to know wikitext anyway to edit what I tried to do). Please drop me a note when VE really works. I won't hold my breath. Fram (talk) 06:58, 24 September 2014 (UTC)

Results of the Victorian state election, 2010 (Legislative Council): Opening the page in VisualEditor takes less than two seconds in Safari and about two seconds in Firefox. It looks like the original in both Safari and Firefox. I had no trouble changing the number of votes that each party received in the template-created table. It's a complex transclusion and works like any other complex transclusion: Find the template you want in the left, choose the parameter you want, and change the value in the field. Moving the box would not have helped you see which parameter feeds which box, because template contents do not auto-update as you type. None of the changes you made would have been visible in the table until after you clicked "apply changes". (Of course, if this were just a table, then you'd have no trouble changing the contents in VisualEditor: click on the number you want to change, and change it. But you couldn't add new rows to such a table in VisualEditor yet, and it would be much uglier to deal with a heavily formatted table in wikitext.) Whatamidoing (WMF)
Obviously, if you test it after someone has corrected the page so that it does work in VE[3], you won't have the same problems. And moving the box would have helped me, to see the match between current (pre-change) field on the screen and in the template... Fram (talk) 19:23, 24 September 2014 (UTC)
Broken tables are broken, regardless of whether they are created with templates or wikitext. Unclosed tables should not be tolerated by the community, because although it looks "correct" on some browsers, it is broken in others and some screen readers. Does it display correctly for you when you open it in VisualEditor now? Whatamidoing (WMF) (talk) 19:32, 24 September 2014 (UTC)
Parameter order is always given in the order that was specified in the TemplateData by an editor here at the English Wikipedia. Anyone who believes that a different order would be preferable is free to re-order the TemplateData into something more sensible. This is entirely under the control of the community. Whatamidoing (WMF)
And what if, like in this case, no template date is given? Oh right, then VE chooses the alphabetical order, not the order already used in the article or the order in the template syntax. Fram (talk) 19:23, 24 September 2014 (UTC)
Yes, if the community chooses not to add TemplateData, then you get the default. Whether to provide non-default order is up to the community. Whatamidoing (WMF) (talk) 19:32, 24 September 2014 (UTC)
And the default is useless. Whether to use the order used in the template, or in the article, is up to the Ve developers. Alphabetical order is mindless and useless. Fram (talk) 19:42, 24 September 2014 (UTC)
The description provided by TemplateData was the choice of an editor here at the English Wikipedia. If you believe that the description is wrong, then you can certainly change it, i.e., by providing individual TemplateData for each infobox instead of re-using the same, one-size-fits-some TemplateData for all of them. This is entirely under the control of the community. Whatamidoing (WMF) (talk) 17:30, 24 September 2014 (UTC)
Right... I added the template data description to the template[4], and it doesn't work. Perhaps I did it wrong, no idea, but the result is what it is... Fram (talk) 19:23, 24 September 2014 (UTC)
Did you purge the template afterwards? Whatamidoing (WMF) (talk) 19:32, 24 September 2014 (UTC)
No, why? I did try it out on another article after changing the template though. Have you purged it and tried it, or are you just asking random questions instead of checking things? Wouldn't be the first time (see this very section, where you blamed template data for a problem, even though the template didn't have templatedata...) Fram (talk) 19:42, 24 September 2014 (UTC)
Maybe you would have done this because you read the directions at Wikipedia:TemplateData/Tutorial#Save, which say that TemplateData needs a null edit to purge the page if you want it to work right away, rather than waiting for whenever the database catches up with your edit? Whatamidoing (WMF) (talk) 20:54, 24 September 2014 (UTC)

So, thank you Whatamidoing (WMF), three answers, not one of them relevant or correct. Can you ask someone from the WMF who actually knows what they are doing (if any can be found) to respond in the future? As this was simply a waste of time. Fram (talk) 19:23, 24 September 2014 (UTC)

Bu, bu, bu ... they're working so HARD...

[Vigilant]

P.S. trimming white space on posting is a stupid default. Please leave white space alone.

[EricBarbour]

VE is like Scientology: the Wiki-gift that keeps on giving. Or polluting.

[Thracia]

Open in VE, takes somewhat long, doesn't look like the original page.

That WYSIWYG editor is really coming along!

[Zoloft]

Open in VE, takes somewhat long, doesn't look like the original page.

That WYSIWYG editor is really coming along!

Unfortunately it's more like this:

[Vigilant]

Balls.
https://meta.wikimedia.org/w/index.php? ... id=9972892

Complaint about some employees of the WMF

As an unhappy contributor, I would like to make a complaint about some WMF employees.

User:Erik Moeller (WMF): he has single-handedly alienated more Wikipedia editors (all languages) than the rest of the WMF combined. I don't think I need to rehash all his problematic decisions, communications, and actions wrt VisualEditor and MediaViewer, which were basically the culmination of years of problems. Please, if you can't or won't get rid of him, then at least make sure that there is no more interaction between Moeller and the projects, and no more communications from Moeller.

User:Jorm (WMF) is the proverbial example of a software designer who is not interested in what people need, only in what he can make. His work on Flow, and his communications about it, emphasized the gap between his vision and what was actually wanted or needed, and his inability to grasp this and act accordingly. Perhaps he is a good developer, but then make sure that he only produces things for which others have written the specifications, and keep him away from interactions with editors.

User:Jdforrester (WMF), doing his best to make the exasperation over VisualEditor worse at every possible moment. Together with Erik Moeller and Jasper Deng (not a WMF employee) the main reason I refuse to edit at Mediawiki any more. A very poor communicator when things go wrong, and the manager of a completely botched product launch. See e.g. en:Wikipedia:VisualEditor/Feedback/Archive 2014 3#Testing report from Wbm1058 for a typical passive-agressive example of his lack of clue (also here).

User:Whatamidoing (WMF) is trying her best, but is just completely incompetent in her work as a community liaison (I go with incompetent, as the alternatives are worse). She is not a community liaison but a crony of the WMF, incapable of seeing problems, incapable of understanding needs, wants, or even basic posts, making up things to defend her (or the WMF's) position, and more often than not a total waste of time. It's always hard to find when things started to go wrong, but where it for me definitely went downhill, never to recover again, was at en:Wikipedia:VisualEditor/Feedback/Archive 2014 1#Feedback request: VisualEditor special character inserter, with unforgettable statements like "I know that some people find the Agile approach irritating, but since you systematically search for incomplete features, you are consenting to it and supporting it. Whatamidoing (WMF) (talk) 23:21, 29 January 2014 (UTC) " Discussions further down that page (e.g. "Empty <ref /> tag added by VE" or "Advanced image settings? More VisualEditor rubbish.") illustrate similar problems with Whatamidoing and JDforrester. These are old, I use them to indicate where things went wrong and for how long these user interaction problems persist already. The current VE feedback page on enwiki shows the ongoing problems with Whatamidoing. See also e.g. en:Wikipedia:VisualEditor/Feedback/Archive 2014 3#Utter waste of money. Another good example is en:Wikipedia:VisualEditor/Feedback/Archive 2014 3#Category additions with VE are FUN!, where Whatamidoingis doing everything but helping. Her attitude is "it is a local enwiki non-VE problem unless you prove to me otherwise", not "hmm, I wonder if this is enwiki only, let's try it out". In many disucssions, she makes categorical statements, which give the impression of her being knowlegeable when they are correct, but only make her look foolish when they are wrong, which happens way too often.

Other editors may have other experiences of course, but for me it really has gone to far for much too long. Fram (talk) 09:31, 25 September 2014 (UTC)