Everything you look for and all that you perceive has a way of proving whatever you believe.

Every now and then I am reminded of cognitive bias as presented by Kruger-Dunning in the paper Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessments. As I have been active in the opensource community for close to 18 years now in some form or fashion the first lesson I learned is that if you have never done it before. You don't know. It's just that simple, and this lesson tends to stick with you with in almost every facet of life. It affects ones approach to problems such that every problem starts with one question. Quickly and summarily followed by a series of questions that need to be answered. My standard trajectory usually begins to branch off in a series of questions leaving me with few answers and even more questions to which I simply don't have the answer. Early on this led to a slight madness as I would start down some rabbit hole and end up working on some esoteric issue that affected a handful of people. In retrospect, I was young and dumb, back then I thought that so long as you plugged away at it. No matter how long it took you, you would be able to answer anything. Now that I'm slightly older, and know how dumb I still am and how valuable it is in choosing the right question, I am able to avoid madness. Of course, sometimes it will sneak up on me and I find myself doing something that makes absolutely no sense at all. This is obviously why we call it madness, but I'm much better at avoiding it or tolerating it depending on my needs or condition. I'm of the mindset that madness can't be cured, only managed.

Usually if I receive an email or a critique of work furnished that needs a sign-off, peer review or some such. I'm wholly open to it because my first assumption is that if I'm treading new or unfamiliar water then I obviously don't know what I am doing. Secondly, the more eyes the better. While treading familiar waters I am much less open to critique or suggestion unless I feel it will beneficial, primarily to avoid rabbit holes but also due to a lack of time. New variables mean risk and if there is no gain by adding the element of risk to something that has been tried and true. I'm unlikely to entertain or even respond to it. This is not necessarily always the case but largely this behavior I feel has strengthened my ability to stay focused on the right question and to avoid rabbit holes.

I think proponents and participants in opensource operate within the same vein and mindset for the most part. It's more of a behavioral mindset or lifestyle, propelled by the some itch, urge, quest to solve a given problem and find the right answer. Whether it be for fortune, fame, enjoyment or some other facet. You work the problem until you are tired and move on to some other problem whether it be in the same domain or somewhere else. Hopefully you were able to answer a question or two and because it's opensource inevitably someone at some point will continue finding answers to questions old or new.

I'd appreciate much more answers however

Lately though with my own interactions I feel as if this isn't the case, partly because I suspect their is a racial and gender tinge to cognitive bias that is ripe for exploration. This brought up in a couple of discussions on gender recently. Specifically in regards to women in opensource and generalizations of women by men in general. Before these discussions a couple of weeks earlier I had recently finished reading a post by Daniel Stenberg titled "Three out of one hundred" (which I recommend you read) so the conversation and ideas were fresh in my head. Also, I had a heated discussion about some work where I can only make the assumption that because of my skin color this person thought that somehow they knew better and would simply do the exact opposite of what I said. To be brutally honest, sometimes this works in my own favor, and to be blunt, I'm a capitalist and love money so cleaning up a mess may work out for me in monetary fashion. I mean, ignorance is costly, but let me be clear. Unlike agile salesmen, I try my best to offer clear and abrupt warning.

On the whole though the idea is that men on average will make the incorrect assumption that they are more capable at a technical subject than their female counterparts no matter how qualified they may be. Women having to deal with this their entire lives tend to see the bigger picture and avoid what they consider to be rabbit holes. As a woman I would ask why? Why even bother going down that route if every time I do someone is going to offer to second guess me? This coalesced quite nicely with another discussion I had on generalizing untrue statements as applied to women. So, let us take for example the statement, "all women are batshit insane" as clearly being a gross generalization. Amongst a group of men shooting the shit  it may be viewed as an off-cuff remark, in a mix of women, it will still be viewed as an off-cuff remark but it will be normally tolerated, at least to your face. However, this one statement in and of itself subtly reaffirms subconsciously that there must be some bit of truth to that statement. Thus, most women are indeed insane, maybe not batshit insane but a mild course of irrational behavior is expected. Of course this statement is simply ridiculous! I don't have the actual data to back any of this up and saying I believe it to be true isn't enough for an idiot, I realize. That aside, I am ashamed to admit that I am guilty of remarks like this myself. I've tried to rationalize the comments with "it's about context" but I can't qualify that as an "african-american" male. Is the statement "black people know jack shit about computers" any different? It's generalistic, and is again ridiculous but do we subconsciously hold onto these ridiculous generalist statements and again subconsciously use them to validate our own cognitive bias? Of course if that is actually the case there will be less women in opensource because all the men seem to think they are superior. Can you really blame someone that can't get a word in edgewise, everything they do is second guessed, or felt to be inferior? Doesn't sound fun to me at all, especially when you are paid less and have to doubly prove yourself answering harder questions. In short, fuck that noise. So of course, there will be less of any group that isn't already in the majority. Which means we have much less answers to our questions. This is unacceptable and we must try harder not just in our professional but personal lives as well.

The Mary Protocol

In response to my own behavior and to help me realize when I may be making a generalist statement that I may not consider hurtful, insulting or a turn-off to someone. I've implemented the Mary protocol. It's a simple procedure that will help me, help myself.

1. The moment I may have said something and/or heard someone else say something that sounds ridiculous enough to either retract with an immediate apology or to pause the conversation with a "not cool". I will do so.
2. I will not argue about my position beyond stating that i'm invoking something known as the Mary protocol to help myself from reinforcing my subconscious with nonsense and to allow for open discussion if the offending party would like to hear more.
3. I will do my best to keep any proceeding conversation succinct and within reason.

The protocol is to be like a bit of a tase to correct my behavior. Not because I want to become some righteous feminist with a penis but because should I have a daughter someday I'm quite sure i'd take high offense to anyone calling her batshit insane, or treating her in inferior manner in anyway shape or form. So much so that I pray to never see it occur in my presence for the safety of the offending party. Off-cuff or not, it's not acceptable behavior and I will not continue it. It is my hope, that you do the same.

Share/Bookmark

# Copyright (c) 2011 Christopher Warner - cwarner@kernelcode.com (http://cwarner.kernelcode.com)
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
#   The above copyright notice and this permission notice shall be included in
#   all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
#
from roundup.anypy.sets_ import set
from roundup import roundupdb, hyperdb

def deptissue(db, cl, nodeid, oldvalues):

    # get all message info
    create = cl.generateCreateNote(nodeid)
    dept_id = cl.get(nodeid, 'department')
    print dept_id

    ''' BUILDING DEPARTMENT '''
    # You have to know the id mapping
    #  technically we could find this

    if dept_id == '1':
        for msgid in cl.get(nodeid, 'messages'):
            try:
                print "Alerting Building Engineering Department"
                cl.send_message(nodeid, msgid, create, ['christopher.warner@drms.dla.mil'])
            except roundupdb.MessageSendError, message:
                raise roundupdb.DetectorError, message

    elif dept_id == '2':
        for msgid in cl.get(nodeid, 'messages'):
            try:
                print "Alerting Technology Department"
                cl.send_message(nodeid, msgid, create, ['Technology EMAIL ADDRESS HERE'])
            except roundupdb.MessageSendError, message:
                raise roundupdb.DetectorError, message

    elif dept_id == '3':
        for msgid in cl.get(nodeid, 'messages'):
            try:
                print "Alerting Research Department"
                cl.send_message(nodeid, msgid, create, ['RESEARCH EMAIL ADDRESS HERE'])
            except roundupdb.MessageSendError, message:
                raise roundupdb.DetectorError, message
    else:
        print "Department doesn't exist"
        pass

def init(db):
    # check on create of ticket
    db.issue.react('create', deptissue)

Share/Bookmark

The idea of a unifying unique identifier or Single Sign-On (SSO) has been around for some time now. Implementors of these systems have put together several different methods of getting this done whether through OpenLDAP, NIS or some other glue. The idea of OpenID implementations itself bears some background however because OpenID is actually a specification and most if not all current implementations follow that specification.

Politics

It's the unspoken verbiage, or simply the bit rot of code but the politics surrounding single sign-on identity is essentially about the same as a Mexican Standoff with a bit of Catch-22. The twist is that the court of public opinion is between the providers, watching, waiting. The idea of allowing the user or customer access to your wealth and dearth of services via their Google id, if you are Yahoo doesn't sound that enticing. However, if one goes out of the way to explicitly disable service you become the legion of bad interoperability, the all of this is mine guy. Yes, that guy. The first one to step out into the foray of public opinion "cutting off access" earns the that glorious badge. We've seen this type of behavior recently between Facebook and Google in regards to contact data. Google didn't feel the quid pro quo there and logically cut that source of information off to Facebook. Even though Facebook was more the proponent of "bad behavior" the court of public opinion was squarely against Google. As in, they had to defend their position. Subsequently we end up with headlines like "Google vs Facebook For Control of Your Reputation" or "Google Vs Facebook: Who's Right & Who's Wrong?" Interestingly enough this was over nothing more than contact data but as we all know, data is a very valuable commodity. So, surely many companies are in competition right now and that competition is good for us all. Even for the purveyors of other online services and venues it is a good thing. That aside for all the versus articles one would think that using your Google ID to sign into Facebook would be forbidden. Thankfully, you would be wrong.

Facebook OpenID Capable - Figure 1

Fortunately, that tidbit is really no secret to anyone actually involved with the single sign-on problem, drafters, implementors, management and all of the in-betweens understand this weird paradoxical stance very well. If we take a look at Figure 1 we can see OpenID clearly works with my Google account; even though we can clearly see Google isn't displayed at all; politics. If you are reading this as a user and are  juggling an average of two or more ids to sign-in to different systems, unfortunately, you are experiencing the problem first-hand. The good news is that OpenID as implemented by most of the major providers does work. That said and let us pass this issue of politics there are also clearly implementation problems that don't make it easy. Which is compounded with the obvious lack of marketing and usability and finally of the functionality and service itself.

The OpenID specification as it currently stands is extremely flexible and the idea of the actual OpenID itself is a solid one. Primarily as a universal unique identifier it leaves authentication to the site or organization you choose to manage your id. It's essentially the panacea that we have all wanted in regards to identification. One unique key, that opens all doors for all the services that we want to use. Unfortunately do to this flexibility we run into implementation problems and this can make the problem a bit more complicated as far as implementation goes.

Usability

The crux of the OpenID issue is primarily the usability piece. The simple fact is that a majority of user are so honed in to bad practice they don't want to change. They want to sign-on as "steve_mcqueen99" and enter their secure password "12345678". 99% of you are saying, well, the password is obviously hilarious (while adding a "change" my password mental note). The other piece is the id "steve_mcqueen99". It's pretty unique, until you realize that steve_mcqueen99 looks something like this "??????99_" and someone in Quy Kim China likes Steve McQueen just as much as you. How do we go about making very unique ids? OpenID uses URI's which is where we hit our first usability issue.

The id is actually a URI (Uniform Resource Identifier) which looks something like this http://profiles.google.com/christopher.warner which in this case also happens to be my unique Openid with Google. The problem with this is that most of our user base simply can't differentiate between a URI and a URL (Uniform Resource Locator) and in all actual fact that URI does indeed link to a URL of my public web profile  at Google.

Can you imagine asking for my id over the phone in  trouble shooting a problem? The conversation would sound something like this.

"Ok sir, let me just have your unique id and I will be able to process that for you"

"Of course, that's h t t p colon slash profiles dot google dot com slash christopher dot warner"

Next up data migration between services, social malaise, pushing auth to the browser and finally into a formal RFC and WG specification.

Share/Bookmark

Specification Pepsi Openindiana Build 148

1. How much energy does the entire machine use?

   On initial boot with the Samsung Optical SH-223L/BEBS spinning up we hit 95 watts, the system then drops to about 50watts and as it idles out drops to between 47 and 48 watts. While pushing the machine using all 4 drives as stated in the spec, we fluctuate between 50 and 57 watts.

2. Why the 750w ATX Power supply?

   Based on the above data the 750w would seem to be overkill but in my specific case I have multiple units plugged into the motherboard. If one doesn't plan on adding more than 4-6 hard drives you could easily get away with a 200-250 watt power supply. I'm going to search around for one and update the original specification.

3. Do you need to have the optical drive?

   Not necessarily, you could technically remove the optical drive after install of the specification if you wanted to save on initial start up power. However the unit spins up only on start-up and considering it will be rarely if ever used, it's pretty much a complete non-issue.

4. No ECC memory, will that be an issue?

   In the original specification I noted the addition of ECC memory. Well, the Intel D5x doesn't support ECC memory most likely to save as much energy as possible. It also doesn't support Dual Channel memory, again for the same reason. Not the biggest deal in the world for this machine as it's primarily used for storage and technically speaking the likelyhood of getting severe degradation or errors are low. Still, it would of been a nice to have. To my knowledge I don't know of any mini-itx motherboard manufacturers that support ECC memory. At least not yet.

5. Instead of the D510, why not the D525?

   Well, if we take a quick look at a comparison between the D510 and the D525 we see that we gain a minute speed bump in processing and DDR3 instead of DDR2 and wattage remains unchanged. None of this quite frankly mattered to me at the time as the D510 was roughly $20.00 USD less. None of the stated improvements would significantly speed up the function of archival and storage. This obviously changes if you plan to do compression or encryption with ZFS in which case it may then become useful. In reality though it would make much more sense to apply that extra $20.00 USD to a proper gzip or encryption off-board card.

6. How much will it cost you to operate this unit?

   I pay roughly 9 cents per kwh off the top of my head. I'll have to update with exacts next time I check but based on that number we are looking at close to $3.05 USD per month with the machine running at an average of 0.47kw/h or roughly $37.00 USD a year. Not such a horrible price for the machine that backs up my lively hood. Not that bad at all.

Share/Bookmark

Specification Pepsi Openindiana Build 148

The machine is named 'pepsi' and consist of the following hardware (all prices are current as of 03/22/2011):

• Lian LI PC-Q08B Black Aluminum Mini-ITX Tower Computer Case - $119.00 USD - Optional
• Intel BOXD525MW Intel Atom D525@ 1.8GHz (Dual Core) BGA559 Intel NM10 Mini ITX Motherboard/CPU Combo- $79.99 USD
• Custom SATA card built to my specifications - $PRICELESS
• Vantec 6-Port SATA II 150 Pci Host Card Model UGT-ST310R - $31.99 USD
• Corsair 2GB 240-Pin DDR2 SDRAM DDR2 800 Model VS2GB800D2 G - $29.99 USD
• Zalman F3 FDB 120mm Fluid Dynamic Bearing Case Fan - $12.99 USD - Optional
• Cooler master R4 Silent 140mm Case Fan R4-S4S-10AK-GP -  $11.99 USD - Optional
• Cooler master GX series 750W ATX Power Supply RS570-ACAAE3-US - $100.00 USD  *
• Samsung DVD+RW Optical drive SH-S223L/BEBS - $28.54 - Optional
• 2x Western Digital 1TB *
• 2x Western Digital 320GB *

* This item is to be part of a refactor meaning that this was originally built for my needs but one doesn't actually need 750 Watts for power and can get by with a 200-250W power supply. Also not included is the price of the 4 hard drives as these are what I had in the original pool. In this case a sufficient comparable HD should do and not fall to far out of bounds in regards to power requirements.

Software wise we have:

• OpenIndiana Build 148 x86

A couple of things to note here on the machine build. The BIOS for the D510 Intel NM10 needed to be upgraded to the latest version as the board I ordered and received was almost a year behind in updates. Newegg did get the delivery to me within 1 day of ordering however which was blazingly fast considering. The fans were ordered because the case comes with a 120 and 140mm fan. The 120mm fan isn't that loud but the 140mm fan has blue leds, which I simply have no tolerance for. The D510 mini-itx board itself is amazingly small and includes a mini pcie port that I can eventually stash a mini pcie wireless card in.

Also, the Vantec 6-Port has 4 internal sata ports and 2 external ESATA ports. It's powered by the SiS 3114 raid controller which is basically some hodge podge mess onboard. They provide a straight-through IDE BIOS upgrade for the chipset available from the Silicon Image manufacturers website. You will need this if you want to use this chipset under OpenIndiana primarily because it doesn't automatically detect this chipset otherwise. Moving from sata to ide may mean no hot-swap but I haven't necessarily tried this yet. I'll update after I hook up another 4TB store, i'm moving the data around now as we speak.

Luckily my custom board wasn't damaged and all of the data on disk were fine. I am currently not at liberty to discuss the custom add-on board but hopefully i'll remember to revisit this post when I am.

Normally I only buy ECC ram however to keep cost low and because I had verified my data wasn't severely damaged, I have foregone the option for this machine. After some burn-in time I will most likely upgrade to an ECC ram module.

Before the hardware failure the most critical and important pool was my research pool which was a RAIDZ 1TB zfs pool. After installing OpenIndiana and setting up netatalk which now supports Time Machine backup I simply did a zfs import -f poolname and was back in business.

Share/Bookmark