Misplaced Pages

User talk:Bbb23: Difference between revisions

Article snapshot taken from Wikipedia with creative commons attribution-sharealike license. Give it a read and then ask your questions in the chat. We can research this topic together.
Browse history interactively← Previous editNext edit →Content deleted Content addedVisualWikitext
Revision as of 01:05, 5 March 2013 view sourceBbb23 (talk | contribs)Autopatrolled, Administrators270,791 edits →Bloomex: re Dimitri and gwickwire← Previous edit Revision as of 09:21, 5 March 2013 view source Daoudamjad (talk | contribs)45 edits →Perfect hash function page cites an algorithm called the CDH algorithm that has 90% similarity to Algorithm II CACM 1992 paper and copyrighted material: new sectionNext edit →
Line 227: Line 227:
*@gwickwire, a couple of comments. I understand you're frustrated, but don't yell at Dimitri or use such strong language ("whatever the hell you want"). Also, telling him how to run his business is inappropriate.--] (]) 01:05, 5 March 2013 (UTC) *@gwickwire, a couple of comments. I understand you're frustrated, but don't yell at Dimitri or use such strong language ("whatever the hell you want"). Also, telling him how to run his business is inappropriate.--] (]) 01:05, 5 March 2013 (UTC)
*--] (]) 01:05, 5 March 2013 (UTC) *--] (]) 01:05, 5 March 2013 (UTC)

== Perfect hash function page cites an algorithm called the CDH algorithm that has 90% similarity to Algorithm II CACM 1992 paper and copyrighted material ==

Bbb23,

My edit is not distruptive. Please note that authors have been contacted for their violation and their response is
posted at "http://iswsa.acm.org/mphf/martin.pdf". and I am posting here the text of their response; Contact me
at daoudamjad@acm.org for details:

Qouting authors of the CDH2009 paper:
"Martin Dietzfelbinger, 2013-02-19
Comparison of two constructions of perfect hash functions
Sources:
Edward A. Fox, Lenwood S. Heath, Qi Fan Chen, Amjad M. Daoud:
Practical Minimal Perfect Hash Functions for Large Databases. Commun.
ACM 35(1): 105-121 (1992)
Djamal Belazzougui, Fabiano C. Botelho, Martin Dietzfelbinger: Hash,
Displace, and Compress. ESA 2009: 682-693
The question is whether Algorithm 1 from is “a copy of” Algorithm 2
from .
I summarize the two algorithms in question, using a similar terminology.
Both deal with a key set S of size n. The purpose is to build a minimal
perfect hash function (in ) or a perfect hash function (), respectively.
The number n is assumed to be arbitrarily large.
In , there is Algorithm 2.
The number r is chosen to be cn/ log
2
n, for some constant c, “typically
smaller than 4” (page 114).
A hash function h1 : U → (assumed to be random) is used to split
S into r buckets. The ith bucket is
Ai = {x ∈ S | h1(x) = i}, for i ∈ .
These buckets are sorted according to size, decreasing.
Then the buckets are “mapped” into a table T of size exactly n, as
follows. Have two random hash functions h0 and h2 with values in .
Treat the buckets one after the other, in the order of decreasing size, down
to size 2, as follows: By trying the numbers in in some random
order s0, s1, . . . , sn−1 , find the smallest j such that the mapping
Ai 3 x 7→ (h0(x) · sj + h2(x) · s
2
j
) mod n
maps the keys in Ai into empty slots in T. As soon as such an sj is found, let
g(i) = sj, mark the corresponding slots as used and go to the next bucket.
1For buckets of size 1, one after the other, find s ∈ such that
g(h1(x)) = s names an empty slot. Let g(i) = s.
Use a bit mark(i) to distinguish buckets of size 1 (mark is set to 0) and
larger (mark is set to 1).
If the construction succeeds, one has a minimal perfect hash function with
constant evaluation time; the space to store the function is r words of length
log
2
n plus r bits (for mark), which gives about cn(1 + 1/ log
2
n)
bits, “typically less than 4n bits” (page 115).
In , there is Algorithm 1.
r is fixed to be n/λ, for some constant λ. The range m is (1 + ε)n for some
constant ε > 0.
A hash function h1 : U → (assumed to be random) is used to split
S into r buckets. The ith bucket is
Ai = {x ∈ S | h1(x) = i}, for i ∈ .
The buckets are sorted by size, decreasing.
It is assumed a sequence (φ1, φ2, φ3, . . .) (in principle arbitrarily long) of
independent fully random hash functions is available at no cost.
Then the buckets are “mapped” into a table T of size exactly n, as
follows. By trying j = 1, 2, 3, . . . one after the other find the smallest j such
that
Ai 3 x 7→ φj(x) mod n
maps the keys in Ai into empty slots in T. As soon as such a j is found, let
σ(i) = j, declare the corresponding slots in T as full.
Transform (σ(i))0≤i<r into compressed form, retaining O(1) access.
It is proved in that with high probability the compressed sequence needs
some constant times n bits. We have a hash function that is 1-1 on S with
range , a factor 1+ε larger than n. The evaluation time is constant.
One can also later apply some known compression tricks to reduce the range
to .
The two algorithms have some similarities. In particular if one says that
x 7→ (h0(x) · sj + h2(x) · s
2
j
) mod n acts like a random function ϕj (which
2is not quite true, but it may be that one doesn’t see it in experiments), the
algorithm in tries up to n random functions until one is found that works
for Ai.
The algorithms differ in some features:
(1) The algorithm in uses range , while the algorithm in uses range
size m = (1 + ε)n. This is a larger range, so the algorithm in is a little
weaker.
(2) The algorithm in uses r = Θ(n/ log n), while the algorithm in
uses r = Θ(n).
(3) The algorithm in stores the value sj for i. There are cn/ log n such
values, each with exactly log n bits, so if c is a constant it is clear that a
linear number of bits is used. The algorithm in stores the index j = g(i)
of a suitable function ϕj for i. These indices usually are very small numbers.
One could store them in binary, in a simple prefix-free code, even with O(1)
access, to get space O(n) bits. With clever compression one can get the space
down to O(n) bits with a small constant. In this respect the algorithm in
makes things more complicated than the algorithm in to reach space
O(n) bits.
It can be stated that the algorithm in is somewhat weaker than the
algorithm in . The perfect hash function has a larger range m = (1+ε)n,
it is assumed that one has fully random functions (φ1, φ2, φ3, . . .), and one
needs some compression algorithm.
There are deeper differences in the behaviour, if one analyzes the algorithms.
In the ESA paper it is proved that the algorithm in has expected
construction time O(n), and if a reasonable compression algorithm is used
(with constant reading time) then the space needed for storing the j-values
for all the buckets is O(n), with a relatively small constant. These proofs
need the larger range m, and they need that r = Θ(n) (and not smaller).
This means that the “weaknesses” (larger range m = (1 + ε)n, random
functions φj) play a central role in that proof, as does compression.
In paper , linear construction time and O(n) bits storage space is claimed
(page 114, third col.). There does not seem to be a mathematical proof to
this effect.
In there are experiments with larger values of r (Table 7, 0.6 words/key,
3meaning r = 0.6n and 0.6n log n bits), where one sees reasonable running
times, looking linear. Further, there are experiments with smaller values of
r (Table 8, value of bits/key growing from 3.60 to 4.58). It seems that em-
pirically, for not too large values of n, the algorithm from finds minimal
perfect hash functions with not too large space. For example: 4.6 bits/key
for n = 4 million in 33,313 seconds. This construction time is high, even
for the machines used in 1990. Figure 10, both parts, gives the impression
that as n increases the running time is linear as long as words/key is a
constant (like 0.2, this corresponds to λ = 5), but if one tries to push the
space to small values (6 bits/key for n = 3.875 million) the construction
time explodes, and for c = 4 no solution can be found at all.
This is in accordance with the following observations regarding the asymp-
totic behaviour of algorithm 2 from . It has the following two properties:
(1) The construction time will grow beyond linear.
(2) For very large n the success probability will be o(1), causing too many
trials with new hash functions, again making linear construction time
impossible.
Remark. Why will Algorithm 2 from still work in the experiments
reported there? The reason is that n = 4 million gives log n ≈ 22, which
together with c = 4 gives r ≈ n/5.5 (with c = 4.6 we have r ≈ n/4.8). In
this range the asymptotics have not yet fully kicked in. It corresponds to λ
around 5 in , and there for ε = 0.01 things still work.
The algorithm in uses a very special design for proving an asymptotic
result. Algorithm 2 from does not have the corresponding properties.
So Algorithm 1 from cannot be “a copy of” Algorithm 2 from .
Explanations of claims (1) and (2) about algorithm follow.
(1) Consider large n. Let λ = (log n)/c. Assume c ≤ 4 is constant. Then
the average bucket size is λ. Consider the period in the placement process
between the time when the table is
1
4
full and
3
4
full. During this period
Θ(n/ log n) buckets are treated, and the size of these buckets is larger than
α log n for some constant α > 0 (with high probability). (This follows from
standard concentration bounds.) If the mapping
Ai 3 x 7→ (h0(x) · sj + h2(x) · s
2
j
) mod n
4is assumed to be fully random, the success probability for one randomly
chosen sj is at most
�3
4
�
|Ai |
≤
�3
4
�
α log n
= n
−δ
,
for some constant δ > 0. This means one has to try n
δ
many sj in ex-
pectation, each test taking time Θ(log n). Since there are Θ(n/ log n) such
buckets, the overall time for this period is Θ(n
1+δ
), superlinear.
It should be noted that this argument works similarly if the “ordering step”
in the algorithm is omitted. The reason is that the average bucket size is
(log n)/c, with most buckets having size between
1
2
(log n)/c and
3
4
(log n)/c.
(2) Consider very large n. Again let λ = (log n)/c. Assume c ≤ 4 is constant.
The load of the buckets is distributed according to the Poisson distribution
with parameter λ. If n is very large, λ will also be large, and asymptotics
of the Poisson distribution will start to hold, such as:
With high probability, 99 percent of the keys will sit in buckets of size
between λ − 3
√
λ and λ + 3
√
λ. (The reason is that the standard deviation
of the Poisson distribution, which governs the bucket sizes, is Θ(
√
λ).) Now
if one looks at the bucket that is treated when the table is 99 percent full, it
will have size larger than
2
3
λ. The probability for a single sj to be successful
is at most
� 1
100
�
|Ai |
≤
� 1
100
�
2
3
λ
≤
� 1
100
�
(log n)/6
.
Since (1/100)
1/6
< .47 <
1
2
, this probability is n
−δ
for some constant δ > 0.
This means that the overall probability that the construction in Algorithm
from succeeds is n
−δ
, so we have to start the whole procedure at least
n
δ
times in expectation. Each such trial takes Θ(n) time."

Revision as of 09:21, 5 March 2013

Archiving icon
Archives
Archive 1Archive 2Archive 3
Archive 4Archive 5Archive 6
Archive 7Archive 8Archive 9
Archive 10Archive 11Archive 12
Archive 13Archive 14Archive 15
Archive 16Archive 17Archive 18
Archive 19Archive 20Archive 21
Archive 22Archive 23Archive 24
Archive 25Archive 26Archive 27
Archive 28Archive 29Archive 30
Archive 31Archive 32Archive 33
Archive 34Archive 35Archive 36
Archive 37Archive 38Archive 39
Archive 40Archive 41Archive 42
Archive 43Archive 44Archive 45
Archive 46Archive 47Archive 48
Archive 49Archive 50Archive 51
Archive 52Archive 53Archive 54
Archive 55Archive 56Archive 57
Archive 58Archive 59Archive 60
Archive 61Archive 62Archive 63

This page has archives. Sections older than 7 days may be automatically archived by Lowercase sigmabot III when more than 10 sections are present.

Caution
  • Unless otherwise requested, I will respond on this page.
  • Please include links to pertinent page(s).
  • Click New section on the top right to start a new topic.

Hansen Site

Would you mind if I unprotect this page? Perhaps you'll remember its appearance at WP:AN3; Heironymous Rowe reported The long road homw for edit-warring for repeatedly attempting to remove it, and you protected it, saying "The other choice was to block both editors. I know diddly about the merits of the content dispute, but battling and screaming at each other is not the way to resolve it." TLRH is edit-warring against standard practise with US archaeology articles (besides H. Rowe's comment about it being a WP:POINT/WP:STALK violation, both of which I've checked and can see to be true) and is basically disrupting things; H. Rowe is trying to maintain article content in the face of disruption, so page protection I believe to be overall unhelpful. Talkback or reply at my talk page, please. Nyttend (talk) 04:20, 25 February 2013 (UTC)

Feel free to do what you think best. Thanks for your courtesy in asking.--Bbb23 (talk) 00:10, 26 February 2013 (UTC)
You're welcome; unprotected, and thanks for your response. The designation that The long road homw was trying to remove is a national designation applied to all US archaeological sites and is thus highly important to include; this is the first time I've ever seen someone object to its presence. Nyttend (talk) 00:24, 26 February 2013 (UTC)
As you noted, when I protected the article, it was a procedural move on my part as I know nothing about the content or what is standard in these sorts of articles. I trust your judgment and leave it in your capable hands.--Bbb23 (talk) 00:36, 26 February 2013 (UTC)

Mohammed Nizamul Huq

Bbb23,

Could I ask you to intervene at Mohammed Nizamul Huq? Edits are being reverted with no good reason. Crtew (talk) 22:11, 25 February 2013 (UTC)

Other than a BLP vio, which I already explained on the talk page hours ago Darkness Shines (talk) 22:13, 25 February 2013 (UTC)
Please double check with the titles I just provided because I'm not sure which is your first or second source. Also are you getting the entire copy from your source?Crtew (talk) 01:05, 26 February 2013 (UTC)

You did an excellent job at reviewing the issues brought up and I pleased with the process. But you've given me more to do ;-) Thank you, Crtew (talk) 01:35, 26 February 2013 (UTC)

I'm glad (about your being pleased). As an overarching issue, you should keep in mind that we should always err on the side of avoiding WP:BLP violations. Thus, if an editor first reverts you, that's when you should discuss the problems on the talk page, not in edit summaries.--Bbb23 (talk) 01:47, 26 February 2013 (UTC)

Jacob Arabo

Hi Bbb23 - I saw his edits also, but this time he did find some decent references and kept the hyperbole to a semi respectable level (see the first efforts!). My repeatedly trained reflex screamed Undo also, but parts do seem to be salvageable. The subject is noteworthy and even notorious in NYC - just last week he got robbed, and it made all the news! I'd hate for his friend's effort to be completely unappreciated. Best--Nixie9✉ 03:14, 26 February 2013 (UTC)

Feel free to salvage what you can, but make sure the sources are reliable (my recollection is they were not) and there's no hyperbole. Thanks.--Bbb23 (talk) 23:54, 26 February 2013 (UTC)

Shallowgravy

I need a check.

The above has a second template - {{Catwoman}} - they'er more than willing to ignore BDR and just bull the rtangental material in be cause "the character (Catwoman) is mentioned in the article".

Yes, there is a discussion, but at this point they are back to the "Argue to undo my bold edit" point.

Since you handled the last one I thought I ask you...

- J Greb (talk) 03:35, 26 February 2013 (UTC)

I'm not sure if I should add my two cents to this or not, but here goes - There has been a line of discussion on the Template talk:Catwoman for some time and after a certain point J Greb stopped adding to the discussion for a few days, so I saw no harm in adding some of what was discussed to the article, as is suggested in Misplaced Pages:BRD, the article was then reverted with no talk or explanation on his part. This has been rather confusing since, once again, it leaves us with no way of moving forward. I would encourage you to take a look at the talk page and, if you could, help me see if I'm getting my point across with civility and perhaps tell me what I might need to do in the future to keep the conversation going and resolve problems.- Shallowgravy (talk) 04:03, 26 February 2013 (UTC)
Yes, of course, you should add your two cents. It helps me understand your perspective. That said, you're essentially repeating the same behavior in this template as you did in the other. When an editor reverts you and a discussion is begun on the talk page, you should continue the discussion in the hopes of reaching an agreement with the other editor (as with the last one, there are only two of you). You haven't convinced J Greb of the soundness of your position. That means your options are the same as they were in other template. You have to get others involved in the process through dispute resolution. As an aside, your comments are perfectly civil. I don't believe that J Greb is accusing you of incivility, perhaps just stubbornness (although that cuts both ways). One more thing related to WP:BRD. In my view, BRD gets a bit confusing when you reach the Discuss and Bold (again) sections. It's not clear, at least not to me, what you're "permitted" to do and what you're not. Remember, as much as it is cited, BRD is an essay, not even a guideline, so it has only so much power. Also, remember above all that edit-warring never solves a dispute and can get you into trouble if you do so. I don't want to block either of you (one can be blocked for edit-warring even without breaching WP:3RR). It may take more time to do it the "right" way, but it has to be done.--Bbb23 (talk) 00:28, 27 February 2013 (UTC)

Isabel dos Santos

Hi Bbb23, I have seen your participation in above article. Thank you. I must advise you that contributor User:Zorglub is editing this article in favour of the corrupt regime there, meaning he eliminates parts of it, that might be negative for the ruling party MPLA and their representatives, so please be careful about what he is doing. Cruks (talk) 08:49, 26 February 2013 (UTC)

User:Nick.mon - edit-warring

Hello Bbb23,

I don't like to be a snitch, but you gave User:Nick.mon a last warning against edit-warring, and he pledged to never do it again, but his resolution has held for only three days. He has done it again: at Five Star Movement, Democratic Party (Italy), The People of Freedom, Left Ecology Freedom. He has rekindled a long ongoing conflict with User:Checco. --RJFF (talk) 21:44, 26 February 2013 (UTC)

Transliteration forms

Transliteration info is available now at Talk:Mohammed Nizamul Huq.Crtew (talk) 22:27, 26 February 2013 (UTC)

Zakir Naik Wikilink

Well, Can you give me a link to an article where they say "no wikilinks in quotes"? As you did in this revision Faizan (talk) 07:11, 27 February 2013 (UTC)

WP:LINKSTYLE ("Items within quotations should not generally be linked").--Bbb23 (talk) 00:41, 28 February 2013 (UTC)

Larry Klayman page

I would appreciate it if you would confirm that the edit I made today to the Larry Klayman page conforms to policy. Thank you.BigDog2012 (talk) 16:56, 27 February 2013 (UTC)

Hi, BigDog, I appreciate your coming here to ask. I reverted your change, but, honestly, it's an unusual case. Normally, we would be considering whether Klayman had ever been convicted of child abuse or some kind of child-related offense. Here, we have a finding by a judge upheld on appeal in a divorce case - and we're reporting that finding. Instinctively, I don't like it. If I had to support my instincts with policy, I would first say that it's not relevant to his notability and therefore shouldn't be included per WP:UNDUE. Second, the implication of the finding is that Klayman may have committed a crime, which means that WP:BLPCRIME applies, which would generally bar the material from inclusion. The finding by the judge is in a civil case and is not held to the same standard of proof as a criminal case. Thus, even though a court made the finding, the finding wasn't subjected to the same scrutiny that would be true in a criminal case.
As in all BLP articles where the issue is not clear, we should err on keeping the information out. That said, if you want to include the material, you can either take it to the article talk page, or you can go directly to WP:BLPN and solicit input from others.
As an aside, please read WP:CITE. The way you cited the newspaper article was not correct.--Bbb23 (talk) 00:59, 28 February 2013 (UTC)

Thanks

for 'unblocking' me.... 8-) Peridon (talk) 21:51, 27 February 2013 (UTC)

Heh, I didn't want to see the imaginary block on your non-existent block log to go unchallenged. So now you have an imaginary unblock rectifying it.--Bbb23 (talk) 00:44, 28 February 2013 (UTC)

Awoods88

Hi Bbb, I don't think Awoods88 (talk · contribs) gets it. Maybe further discussion is merited. Cheers, from 99 on vacation, sort of. 166.147.120.151 (talk) 23:16, 27 February 2013 (UTC)

I'm not sure if I should help you until you revisit your priorities and discuss Downton Abbey with me. Harumph. And what does "sort of" on vacation mean (if you can say)?
On the less important Misplaced Pages issue, it would help if you would pinpoint what Awoods88 is doing wrong. Since he was warned by me on Valentine's Day, he's only made a handful of article edits, all to Don Carmody, part of which were reverted based on copyright violation. Nor has he created any new inappropriate articles. The main thing he's done is engage in a content discussion at Talk:XMG Studio with you and Qworty. Your last comment there seems to imply that Awoods has made "ltierally dozens" of inappropriate article edits since he was warned ("fresh start"). Putting aside the merits of the content dispute, that doesn't seem quite accurate. Am I missing something?--Bbb23 (talk) 01:10, 28 February 2013 (UTC)
Business first: you're right. I was referring to their 'old' history, and interpreting the recent discussion as a continued attempt to find a way of restoring unsourced and/or promotional content, i.e. respecting the rule of your warning, while nibbling away at the same bone. Frankly, my take is that they're incapable of writing an original sentence that isn't promotional, even at Carmody, which I undid some of. But that's me.
On working vacation, but away from my 99 keyboard. Sickened by the final episode of the Kennedys, er, Downtown. Jesus, do they have to kill off all the young ones? The pundits say that Mary will be more interesting now that she's single, but I don't buy that Matthew would have been so careless. Meh. 166.147.120.172 (talk) 01:30, 28 February 2013 (UTC)
The ultimate outcome of Awoods may be inevitable, but right now I'm not prepared to intervene.
I believe they're killing off those actors who don't want to appear anymore. I don't care about Matthew much; I was tired of the character. The whole finale, though, was a bit much, although it's hard to say anymore what counts as "a bit much" in what increasingly is drivel. At least the costumes are still fun. Does any woman appear in the same outfit more than once, or even wear the same jewelry?
Better hope Drmies doesn't catch you using "meh".
Hope you're vacationing more than you're working and that wherever you are is more interesting than home.--Bbb23 (talk) 01:37, 28 February 2013 (UTC)
Drmies drinks ales I never heard of; I think he might have a home brewery going. They do moonshine down there, after all.
I'm a sap to take the Downton story line to heart. It's silly, but a truffle of a pleasure.
Visiting family, working this weekend. Miss my gal and my dogs, but love the people I'm with. 166.147.120.145 (talk) 01:45, 28 February 2013 (UTC)

5 year watchdog

FWIW, my "watchdog" has had me in his sights for nigh on 5 years now <g> even though I avoid going to any pages he edits. My original User:Ikip run-in was about the hundreds of "bilateral relation" AfDs ... but he found a bunch of editors to notify with a hundred or so edits to join his crusade (some of which indeed turned out to be socks of his or of others). Including some "Misplaced Pages Review" types as well, and some real characters (one using "dildo" in his name). Cheers. Collect (talk) 12:05, 1 March 2013 (UTC)

Escalation

I'm sorry to bother you again, and I am now aware of the canvassing policy (I wasn't before), but things are escalating across multiple articles due to the efforts of the three editors I mentioned (Msoamu, Shabiha and Hassanfarooqi). I am now clearly being accused of belonging to a "terrorist organization," in addition to still constantly being ascribed to ideologies (namely Salafism and Wahhabism) which I have expressly said I don't agree with and fear personal/professional repercussions should my IP address ever be tied to such accusations. The timing of all these accusations really leads me to believe that this is not only coordinated among them, but specifically in reaction to my personal concerns. I would really like something done about this as soon as possible, whatever it is. I don't want to feel that I am putting myself at risk simply by logging in to Misplaced Pages. Some of the info is at WP:ANI. MezzoMezzo (talk) 17:46, 1 March 2013 (UTC)

For the purposes of full disclosure, I posted similar messages on the talk pages of User:In ictu oculi, User:Lukeno94, User:Qwyrxian and User:Jimfbleak. I did so because I feel part of the issue is the exact timing that all this is occurring, and would like to increase chances of someone intervening as soon as possible. Hassanfarooqi made a clear threat in one of his comments as shown at ANI and I don't want to find out what that threat will be. MezzoMezzo (talk) 17:55, 1 March 2013 (UTC)
Which is, of course, the veritable definition of canvassing ... (✉→BWilkins←✎) 21:48, 1 March 2013 (UTC)
Mezzo, putting aside the canvassing issue, I have a similar reaction to Qwyxrian's in his tentative response on his talk page (so many places, Mezzo, it's a bit much): do you have some diffs to support your "escalation" ("terrorist organization")? I don't want to go hunting around for them.--Bbb23 (talk) 23:32, 1 March 2013 (UTC)
He heavily insinuated that I am a part of some "terror organization" here, and then made a vague threat here. I apologize if this looks like canvassing, but I am receiving threats and being accused of very serious things, all in a short span of time; what else am I supposed to do? Nobody seems to help unless I act out like this, and I fear that if I don't act out and things like this continue, my IP address' association with such keywords could have real-world repercussions. I don't want to specify what those are but suffice to say that, for the first time in almost seven years of editing, I literally fear that using Misplaced Pages could cause real-world problems for me. That's why I posted on "the talk pages of concerned editors," per the "Appropriate notification" section at Misplaced Pages:Canvassing. MezzoMezzo (talk) 03:38, 2 March 2013 (UTC)
So he said your edits make it look like you have a POV, and then CLEARLY told you to stay of his talkpage or he'll retaliate: that means don't edit with a POV, and avoid his talkpage like the plague. That's not a "vague threat", it's a pretty clear "leave me alone" threat. How in the world could he have your name put on some list - unlike some of us, you don't use your real name at all. Of course, him asking you to give your real name might violate WP:OUTING, but you're doing youself no favours by feeding him (✉→BWilkins←✎) 11:12, 2 March 2013 (UTC)
After reeding DFTT, I think I get what you mean about the "feeding" issue. I nominated some articles he created for AfD, and was under the impression that informing him with a template was a rule. Regarding my name on a list, I didn't mean to insinuate the user would do it. I'm talking about my IP address being on some watch list due to some of the specific accusations I mentioned at WP:ANI; I guess it's my fault for not being clearer. There is some more detail there, which I didn't mention here, though perhaps I should have; the ordeal has upset me somewhat and I suppose I didn't make the best choices. Regardless, I do stand by my lack of comfort with the user and the accusations at this point, hence my asking for help. Again, I'm sorry for any trouble from my direction. MezzoMezzo (talk) 12:43, 2 March 2013 (UTC)
How would he get your IP address? That information is not accessible to him. Normally, you *should* notify them of AFD's, etc, although in cases where he has specifically asked you to leave him alone, it becomes a bit messy. The more prudent thing to do might be to have someone else either AFD it and/or notify him so that it doesn't look like you're merely following him around nominating everything he does for deletion - that would tend to cheese someone off (✉→BWilkins←✎) 13:18, 2 March 2013 (UTC)
Mezzo, I agree with some of what BWilkins says but not all of it. There are an awful lot of issues that are being mixed up here, and it makes it hard to give you a coherent answer. At this point, Hassanfrooqi has been blocked and hasn't said anything since being blocked, i.e., complained or requested an unblock (he esssentially invited the block anyway). The ANI discussion is still outstanding, and I don't know if I or another admin will take any additional action on the proposal to warn all editors, assuming no one comes up with an alternative proposal. Only you an evaluate your real-life issues, not us, and as Qwyrxian said on his talk page, your real life is more important than Misplaced Pages. BWilkins is right, though; normally, a user has no acccess to another user's IP address(es) unless the user somehow connects them to himself. The easiest way to avoid that is never edit Misplaced Pages unless you're logged into your account.--Bbb23 (talk) 14:08, 2 March 2013 (UTC)
  • It's definitely a complex issue, with deep roots in the past. With regards to canvassing, I'm keeping an eye on the discussion myself, so any notification about new events just makes me go back and check again. Lukeno94 (talk) 22:49, 2 March 2013 (UTC)
Thank you all for your patience. I've calmed down, and it does seem that I overreacted to the personal comments. It seems I have a lot to think about...at the minimum, perhaps a Wikibreak is in order after the issue is resolved. I need to sleep on this for a bit. MezzoMezzo (talk) 03:48, 3 March 2013 (UTC)

Possibly interesting

Misplaced Pages:Administrators'_noticeboard/Incidents#User:Polllilur - now editing as User:Baeer. Peridon (talk) 17:49, 2 March 2013 (UTC)

Thanks for the heads up. I commented there.--Bbb23 (talk) 18:08, 2 March 2013 (UTC)

Deleted Sandbox

I would like to ask that a deleted work (deleted by Hal9000W) be recovered so I can fix problems with the document. Thanks. — Preceding unsigned comment added by Stealth57 (talk • contribs) 19:52, 2 March 2013 (UTC)

Your request makes no sense, but I assume you're talking about the article of yours I deleted. It cannot be restored as it was a copyright infringement.--Bbb23 (talk) 19:58, 2 March 2013 (UTC)

Hassanfarooqi

Help Needed

I was dragged into ANI by MezzoMezzo.I have not lodged a complaint about him.It was between him and Msoamu.I dont know canvassing and did not do it ,I did not troll either.Iam simply creating Articles and trying to edit them with NPOV.I have been called Barelvi Barelvi many times, a terminology disliked by Sufis.PA are common on me.I have not replied in the same tone.I have seen incitements and then reactions in case of other editors.How can I save from these disputes. Shabiha (talk) 08:32, 3 March 2013 (UTC)

It's worth mentioning that Shabiha entered himself into the ANI discussion with this edit, a day before I mentioned Shabiha. He wasn't dragged into anything. MezzoMezzo (talk) 09:25, 3 March 2013 (UTC)

Issue not solved

"There's no reason to leave this open for Strangesad to use as a soapbox or to further discuss the content issue." What I said about the content issue was a response to what others--mostly admins--said about it to me. Strangesad (talk) 18:07, 3 March 2013 (UTC)

Vodka

Might be worth it to leave Poeticbent's 3RR report on the board because there were at least two IPs doing the same revert earlier in February. If they come back my guess is that semiprotection will be needed. EdJohnston (talk) 22:57, 3 March 2013 (UTC)

I restored it and put in the result. Are we using ANEW as a to-do list? :-) --Bbb23 (talk) 23:39, 3 March 2013 (UTC)
If what nation invented hummus is an ARBPIA issue then vodka deserves to be in ARBEE. Though of course we hope it is all settled now. EdJohnston (talk) 00:07, 4 March 2013 (UTC)
God help you if you drink vodka while eating hummus.--Bbb23 (talk) 00:10, 4 March 2013 (UTC)

Your efforts

The Barnstar of Diligence
For the time you put into picking apart messy conflicts and still being objective when assessing error, even (or especially, as you helped me in more ways than one) when some of those errors happen to be mine. MezzoMezzo (talk) 03:26, 4 March 2013 (UTC)
Thanks, Mezzo.--Bbb23 (talk) 00:31, 5 March 2013 (UTC)

Thank You

I am much obliged for your neutral and objective decision in this case.Thank You very much.I request you to kindly watch some Islam related pages like Barelvi Sufi Salafi Wahabi if you have time. Shabiha (talk) 07:12, 4 March 2013 (UTC)

Thanks, Shabiha, if you believe any of the involved editors have violated the warnings issued by Qwyxrian, feel free to let me know. Also, if you need some guidance in avoiding violating your own warning, feel free to let me know.--Bbb23 (talk) 00:33, 5 March 2013 (UTC)

Requesting article edit

I have agreed not to edit the article for a period of time, but this edit should probably be removed before then. -- TRPoD aka The Red Pen of Doom 17:04, 4 March 2013 (UTC)

Notice of Dispute resolution discussion

Hello. This message is being sent to inform you that there is currently a discussion at Misplaced Pages:Dispute resolution noticeboard regarding a content dispute in which you may have been involved. Content disputes can hold up article development, therefore we are requesting your participation to help find a resolution.

Guide for participants

If you wish to open a DR/N filing, click the "Request dispute resolution" button below this guide or go to Misplaced Pages:Dispute resolution noticeboard/request for an easy to follow, step by step request form.

What this noticeboard is:
  • It is an early step to resolve content disputes after talk page discussions have stalled. If it's something we can't help you with, or is too complex to resolve here, our volunteers will point you in the right direction.
What this noticeboard is not:
  • It is not a place to deal with the behavior of other editors. We deal with disputes about article content, not disputes about user conduct.
  • It is not a place to discuss disputes that are already under discussion at other dispute resolution forums.
  • It is not a substitute for the talk pages: the dispute must have been discussed extensively on a talk page (not just through edit summaries) before resorting to DRN.
  • It is not a court with judges or arbitrators that issue binding decisions: we focus on resolving disputes through consensus, compromise, and explanation of policy.
Things to remember:
  • Discussions should be civil, calm, concise, neutral, and objective. Comment only about the article's content, not the other editors. Participants who go off-topic or become uncivil may be asked to leave the discussion.
  • Let the other editors know about the discussion by posting {{subst:drn-notice}} on their user talk page.
  • Sign and date your posts with four tildes "~~~~".
  • If you ever need any help, ask one of our volunteers, who will help you as best as they can. You may also wish to read through the FAQ page located here and on the DR/N talkpage.

Please take a moment to review the simple guide and join the discussion. Thank you!

Bloomex

Hello Bbb23, I am owner of Bloomex, the floral company operating in Canada, Australia and USA. We employ 200+ people and do quite a volume in floral deliveries. We have an article on Misplaced Pages, which does not reflect the truth about company or operations mainly due to efforts of CliffC and bots he creates. Any success to combat him are ineffecgtive due to his experience as Misplaced Pages editor and obvious luck of interest from Misplaced Pages public to Bloomex. The main issue is Controvercy page. In my opinion it has to be assessed WP:UNDUE and by then removing that long and deformative text maintained by CliffC. I have put my version as "During its growth stage company attaracted certain amount of customers complains. It was reflected in Toronto Star and CBC Marketplace "

You are very trusted and experienced editor and I would like to ask your help in either removing article completely or leaving it with set of true facts reflecting many years of work, customers we served and dedication and hard work of employees. Sincerely Dimitri Lokhonia Dimitri Lokhonia (talk) 19:40, 4 March 2013 (UTC)

Dimitri. Before you go yelling that what I'm saying here is rude, don't waste your time:
You're the owner. You have a conflict of interest. You've never ever read that page. If you have, you must not have comprehended it.
Misplaced Pages does not care about truth. Misplaced Pages cares about verifiablility of content. UNLESS you can provide sources that explicitly say that the content we have is wrong, then you will not get it removed at all. It is not undue weight. Undue weight would be if 10% of sources covered the controversy, and the rest (90%) covered nothing of the sorts. Fact is, 90% of sources are covering your company's controversies, and there are many. So it makes sense that our article mostly covers the controversy. Your version of "certain amount of complains" first of all is not grammatically correct. Secondly, it would be not giving it due weight, as the vast majority of sources cover the controversy. The two sources you provided go to prove you are wrong. You provided two sources that cover the controversy. Since even you can't provide any sources that don't cover the controversy, then we will continue to have it in the article until such time as you can prove that a majority of sources do not cover your controversy. I'm sorry that this may be hurting your business. But maybe instead of trying to erase it from the history books (NOT what Misplaced Pages is for, we are for recording history) you should try to learn from the controversy and then fix it. Then maybe when your company actually gets covered in a good light in newspapers can we change the current lean of the article. gwickwireediting 21:12, 4 March 2013 (UTC)

Dear gwickwire and CliffC, I will appreciate if you leave me and company alone and will not participate in edit warring. I know you point of view, but unfortunately it does not reflect the true picture. Let other people decide about the correct content of the article Dimitri Lokhonia (talk) 22:49, 4 March 2013 (UTC)

You edit warred against consensus. Just because you own the company doesn't mean we do whatever the hell you want. Especially because you're now doing what's called Forum Shopping (Here, another talkpage, talkpage of article, DRN, COI noticeboard) for a good response. YOU WILL NOT GET A RESPONSE YOU WANT. Your company made mistakes, they were reported on in reliable sources. We report on what's in reliable sources. Therefore, we report on the mistakes. Sorry. Not our problem now. Fix your company, get some new newspaper articles about you, then come back. I'd like to apologize to Bbb23 now that this had to come to his talkpage. gwickwireediting 23:25, 4 March 2013 (UTC)
  • Dimitri, here are my comments, in no particular order:
  1. Don't accuse editors of sock puppetry as a means of getting what you want. It can be perceived a personal attack and can be grounds for being blocked. If you have solid evidence of sock puppetry, feel free to report it at WP:SPI.
  2. The article is locked until March 9. It is unlikely that any content in it will be changed during the lock.
  3. As the owner of the company, you should stop making changes to the article, particularly the kinds of changes you've made. If you have suggestions, you'll have to restrict yourself to the talk page.
  4. If you think the article should be deleted, once the article is unlocked, you can nominate it for deletion. Remember, though, that articles are deleted because they don't meet Misplaced Pages's notability guidelines, not because the company or the owner of the company doesn't like what the article says. Also, bear in mind that it has been through two previous deletion discussions, one in 2009 and the most recent in 2010. Usually, something has to have significantly changed to renominate it for deletion. I didn't look at the state of the article in 2009, but it doesn't look like too much has changed since the 2010 discussion.

Perfect hash function page cites an algorithm called the CDH algorithm that has 90% similarity to Algorithm II CACM 1992 paper and copyrighted material

Bbb23,

My edit is not distruptive. Please note that authors have been contacted for their violation and their response is posted at "http://iswsa.acm.org/mphf/martin.pdf". and I am posting here the text of their response; Contact me at daoudamjad@acm.org for details:

Qouting authors of the CDH2009 paper: "Martin Dietzfelbinger, 2013-02-19 Comparison of two constructions of perfect hash functions Sources: Edward A. Fox, Lenwood S. Heath, Qi Fan Chen, Amjad M. Daoud: Practical Minimal Perfect Hash Functions for Large Databases. Commun. ACM 35(1): 105-121 (1992) Djamal Belazzougui, Fabiano C. Botelho, Martin Dietzfelbinger: Hash, Displace, and Compress. ESA 2009: 682-693 The question is whether Algorithm 1 from is “a copy of” Algorithm 2 from . I summarize the two algorithms in question, using a similar terminology. Both deal with a key set S of size n. The purpose is to build a minimal perfect hash function (in ) or a perfect hash function (), respectively. The number n is assumed to be arbitrarily large. In , there is Algorithm 2. The number r is chosen to be cn/ log 2 n, for some constant c, “typically smaller than 4” (page 114). A hash function h1 : U → (assumed to be random) is used to split S into r buckets. The ith bucket is Ai = {x ∈ S | h1(x) = i}, for i ∈ . These buckets are sorted according to size, decreasing. Then the buckets are “mapped” into a table T of size exactly n, as follows. Have two random hash functions h0 and h2 with values in . Treat the buckets one after the other, in the order of decreasing size, down to size 2, as follows: By trying the numbers in in some random order s0, s1, . . . , sn−1 , find the smallest j such that the mapping Ai 3 x 7→ (h0(x) · sj + h2(x) · s 2 j ) mod n maps the keys in Ai into empty slots in T. As soon as such an sj is found, let g(i) = sj, mark the corresponding slots as used and go to the next bucket. 1For buckets of size 1, one after the other, find s ∈ such that g(h1(x)) = s names an empty slot. Let g(i) = s. Use a bit mark(i) to distinguish buckets of size 1 (mark is set to 0) and larger (mark is set to 1). If the construction succeeds, one has a minimal perfect hash function with constant evaluation time; the space to store the function is r words of length log 2 n plus r bits (for mark), which gives about cn(1 + 1/ log 2 n) bits, “typically less than 4n bits” (page 115). In , there is Algorithm 1. r is fixed to be n/λ, for some constant λ. The range m is (1 + ε)n for some constant ε > 0. A hash function h1 : U → (assumed to be random) is used to split S into r buckets. The ith bucket is Ai = {x ∈ S | h1(x) = i}, for i ∈ . The buckets are sorted by size, decreasing. It is assumed a sequence (φ1, φ2, φ3, . . .) (in principle arbitrarily long) of independent fully random hash functions is available at no cost. Then the buckets are “mapped” into a table T of size exactly n, as follows. By trying j = 1, 2, 3, . . . one after the other find the smallest j such that Ai 3 x 7→ φj(x) mod n maps the keys in Ai into empty slots in T. As soon as such a j is found, let σ(i) = j, declare the corresponding slots in T as full. Transform (σ(i))0≤i<r into compressed form, retaining O(1) access. It is proved in that with high probability the compressed sequence needs some constant times n bits. We have a hash function that is 1-1 on S with range , a factor 1+ε larger than n. The evaluation time is constant. One can also later apply some known compression tricks to reduce the range to . The two algorithms have some similarities. In particular if one says that x 7→ (h0(x) · sj + h2(x) · s 2 j ) mod n acts like a random function ϕj (which 2is not quite true, but it may be that one doesn’t see it in experiments), the algorithm in tries up to n random functions until one is found that works for Ai. The algorithms differ in some features: (1) The algorithm in uses range , while the algorithm in uses range size m = (1 + ε)n. This is a larger range, so the algorithm in is a little weaker. (2) The algorithm in uses r = Θ(n/ log n), while the algorithm in uses r = Θ(n). (3) The algorithm in stores the value sj for i. There are cn/ log n such values, each with exactly log n bits, so if c is a constant it is clear that a linear number of bits is used. The algorithm in stores the index j = g(i) of a suitable function ϕj for i. These indices usually are very small numbers. One could store them in binary, in a simple prefix-free code, even with O(1) access, to get space O(n) bits. With clever compression one can get the space down to O(n) bits with a small constant. In this respect the algorithm in makes things more complicated than the algorithm in to reach space O(n) bits. It can be stated that the algorithm in is somewhat weaker than the algorithm in . The perfect hash function has a larger range m = (1+ε)n, it is assumed that one has fully random functions (φ1, φ2, φ3, . . .), and one needs some compression algorithm. There are deeper differences in the behaviour, if one analyzes the algorithms. In the ESA paper it is proved that the algorithm in has expected construction time O(n), and if a reasonable compression algorithm is used (with constant reading time) then the space needed for storing the j-values for all the buckets is O(n), with a relatively small constant. These proofs need the larger range m, and they need that r = Θ(n) (and not smaller). This means that the “weaknesses” (larger range m = (1 + ε)n, random functions φj) play a central role in that proof, as does compression. In paper , linear construction time and O(n) bits storage space is claimed (page 114, third col.). There does not seem to be a mathematical proof to this effect. In there are experiments with larger values of r (Table 7, 0.6 words/key, 3meaning r = 0.6n and 0.6n log n bits), where one sees reasonable running times, looking linear. Further, there are experiments with smaller values of r (Table 8, value of bits/key growing from 3.60 to 4.58). It seems that em- pirically, for not too large values of n, the algorithm from finds minimal perfect hash functions with not too large space. For example: 4.6 bits/key for n = 4 million in 33,313 seconds. This construction time is high, even for the machines used in 1990. Figure 10, both parts, gives the impression that as n increases the running time is linear as long as words/key is a constant (like 0.2, this corresponds to λ = 5), but if one tries to push the space to small values (6 bits/key for n = 3.875 million) the construction time explodes, and for c = 4 no solution can be found at all. This is in accordance with the following observations regarding the asymp- totic behaviour of algorithm 2 from . It has the following two properties: (1) The construction time will grow beyond linear. (2) For very large n the success probability will be o(1), causing too many trials with new hash functions, again making linear construction time impossible. Remark. Why will Algorithm 2 from still work in the experiments reported there? The reason is that n = 4 million gives log n ≈ 22, which together with c = 4 gives r ≈ n/5.5 (with c = 4.6 we have r ≈ n/4.8). In this range the asymptotics have not yet fully kicked in. It corresponds to λ around 5 in , and there for ε = 0.01 things still work. The algorithm in uses a very special design for proving an asymptotic result. Algorithm 2 from does not have the corresponding properties. So Algorithm 1 from cannot be “a copy of” Algorithm 2 from . Explanations of claims (1) and (2) about algorithm follow. (1) Consider large n. Let λ = (log n)/c. Assume c ≤ 4 is constant. Then the average bucket size is λ. Consider the period in the placement process between the time when the table is 1 4 full and 3 4 full. During this period Θ(n/ log n) buckets are treated, and the size of these buckets is larger than α log n for some constant α > 0 (with high probability). (This follows from standard concentration bounds.) If the mapping Ai 3 x 7→ (h0(x) · sj + h2(x) · s 2 j ) mod n 4is assumed to be fully random, the success probability for one randomly chosen sj is at most �3 4 �

≤ �3 4 � α log n = n −δ , for some constant δ > 0. This means one has to try n δ many sj in ex- pectation, each test taking time Θ(log n). Since there are Θ(n/ log n) such buckets, the overall time for this period is Θ(n 1+δ ), superlinear. It should be noted that this argument works similarly if the “ordering step” in the algorithm is omitted. The reason is that the average bucket size is (log n)/c, with most buckets having size between 1 2 (log n)/c and 3 4 (log n)/c. (2) Consider very large n. Again let λ = (log n)/c. Assume c ≤ 4 is constant. The load of the buckets is distributed according to the Poisson distribution with parameter λ. If n is very large, λ will also be large, and asymptotics of the Poisson distribution will start to hold, such as: With high probability, 99 percent of the keys will sit in buckets of size between λ − 3 √ λ and λ + 3 √ λ. (The reason is that the standard deviation of the Poisson distribution, which governs the bucket sizes, is Θ( √ λ).) Now if one looks at the bucket that is treated when the table is 99 percent full, it will have size larger than 2 3 λ. The probability for a single sj to be successful is at most � 1 100 �

≤ � 1 100 � 2 3 λ ≤ � 1 100 � (log n)/6 . Since (1/100) 1/6 < .47 < 1 2 , this probability is n −δ for some constant δ > 0. This means that the overall probability that the construction in Algorithm from succeeds is n −δ , so we have to start the whole procedure at least n δ times in expectation. Each such trial takes Θ(n) time."

  1. Toronto Star, "Same price, but fewer tax returns" (second section), March 8, 2008
  2. CBC News, "Online flower shop fails to deliver", April 9, 2010