Musings from kb8ojh.net

In a discussion with an academic colleague of mine, the topic of references (in the sense of citations) came up. I’m always nit-picky about references and reference formatting, and I get cranky about bad references — particularly when a publication is under the gun to make a deadline. I haven’t always been; in fact, I’m quite sure I was a burden to my thesis advisor in this department on several occasions. My colleague asked why I care so much “on an initial submission” (as in a submission for review, rather than a camera-ready copy). I hadn’t really expressed my reasoning before, but I did, and here it is again.

It’s about professionalism and attention to detail. While every part of every publication should be correct, complete, and properly laid out, some parts are more important than others. Those parts are the parts that people who want to know what your work is about, and get a feel for whether they should care, are most likely to look at. That is: the abstract, introduction, conclusions, and references. Those four parts of any publication are going to be visited by more eyeballs than all of the painstaking work you presented in the parts in between. Those are also the parts that a reviewer is going to glance over first, before setting in her mind what she expects out of your paper and starting a serious review. It’s important to make a good first impression.

I don’t have any numbers (although I’d be surprised if they aren’t available for at least some fields), but I am thoroughly convinced that typos, errors of grammar, and formatting faux pas in the introduction of a paper are a quicker route to the chopping block than minor technical errors in the meat of the paper. They raise in the mind of the reviewer an image of a scientist who just didn’t care enough about his work to pay attention to the details of presentation. Typos and errors in the references aren’t quite as bad, but they’re right up there. In particular, a typo or error in a citation for a paper written by the reviewer could be quite the blunder.

I don’t get too fussy about citation styles. Various journals, conferences, professional societies, and whoever else specifies such things have different preferences for what each citation should look like. What I do get fussy about is internal consistency. If you use last names but first initials on some of your references and full first names on others, it looks sloppy. Same if you abbreviate venues aggressively in some entries and not others, or include conference location for only certain venues.

This is not to even get started on authors with names like “Märten”. That sort of copy-and-paste error is simply an abomination.

So … check your references. Don’t just check them to make sure they represent what you want to cite, as the saying is usually used, check them to make sure they look good, are consistent, and are typographically correct. I can’t promise it will lead to greater paper acceptance rates, but I can promise that there are people who will notice and appreciate it, and say to themselves, “She really has her ducks in a row.”

We just got back from a few weeks in Europe. On our way out of the country, we were detained for additional screening at the TSA[1] security checkpoint at JFK. In the process, we left behind our laptops. This is the story of how it happened, where the process failed, and how we (ultimately) got them back.

Our flight itinerary took us through JFK and then overseas, but our final domestic leg on the way to JFK was greatly delayed due to some problem “on the ground” at JFK that prevented us from taking off at the origin. We never heard exactly what the problem is, and it doesn’t matter. This delay caused us to miss our overseas flight, but we were reserved seats on another flight the same day to the same destination. Unfortunately, as we have often encountered in the past, our reservations (which were a codeshare on a different airline) were not processed correctly, and it took over an hour to get our boarding passes straightened out. This left us passing through the TSA security checkpoint approximately one hour before our scheduled departure. International flights can (but seldom do) close their doors long before scheduled takeoff, so this was tighter than we were comfortable with.

As our bags went through the security scanner, a TSA agent held up my daughter’s diaper bag and asked whose bag it was. I replied that it was mine, and she asked me to step aside with her. They had decided that a small sippy cup with some milk in it was clearly dangerous contraband, and possibly explosive, and that it needed to be examined carefully. (I know this is standard procedure, but it doesn’t make it any less ridiculous. As if liquids are somehow magically more dangerous than solids.) Unfortunately, the chemical sniffer was out of test strips, so she had to depart the station to find one. This didn’t take very long, but it did take a couple of minutes. She returned, sniffed the milk, ascertained that it was in fact a child’s milk (or at least not an explosive compound), and allowed us to proceed.

This was annoying and stupid, but not really a huge problem in the big scheme of things — Delta’s bungling of our ticket transfer was a lot worse — except that in the hassle and confusion of being pulled aside, we grabbed all of our bags and possessions except our laptops, which had been stupidly removed from our bags in yet another aspect of the useless security theater put on by the TSA (examples here, here, here, and of course here, and those are just the first four I found in a quick search).

No one attempted to find us, no announcements were made of left baggage at the TSA checkpoint, nothing. We simply boarded our plane and flew away for three weeks, unknowing of the fact that we had left our laptops — and our data — in the hands of the TSA. To be fair, identifying us specifically from the laptops would have been difficult, as they are not externally marked; Marina’s laptop does, however, have her name visible on the login screen when it boots. They could have at least tried.

I contacted the TSA at my earliest opportunity from overseas, when we found the laptops missing and figured out what must have happened. The only contact point they give is a phone number, which is very inconvenient for international travelers. It does, at least, give an email address in its answering message, before it drops you to a voice mail box with no opportunity to ever speak to a human being. I sent an email to that email address immediately (rather than paying through the nose to leave a message on the cell phone from which I was calling) describing our laptops and asking what I should do to get them back.

I got no reply in the next day, even accounting for time zone differences. So I called back (by this time I’d established a way to call that only cost me about 25 cents per minute) and left essentially the same information with both my email address and a callback number with voicemail that I could check online. Again, no reply for a day or more. I emailed again, a shorter message with little more than contact information and a brief description of the laptops (in case the TSA agent at the other end just didn’t have the patience and/or literacy to read through my detailed description from before). No reply to that, either. So I called again, and left a very terse and somewhat annoyed message. About a day after that, I got a reply to my second email.

The reply to my second email showed that nobody had read my first email (or at least, not connected the two), because it asked for some of the information from that email. Essentially, they said that they had some items that might be the property I was describing, but that I would have to give them identifying information for the lost laptops so that they could verify. At that point, and only at that point, would they let me know what my options were. Fortunately I was able to determine uniquely identifying information for both machines, and after another annoying delay the TSA confirmed that they had my laptops in their possession and that I might be able to actually get them back. More than a week had gone by, so sending them to catch up with us was no longer reasonable, but sending them home looked plausible.

For other people who might be in this position, the process for getting lost items back from the TSA (at least, from the TSA screening areas at JFK, which may not be consistent across airports) is as follows:

1. Identify your property by general description and specific identifying information (such as serial number, property tag, information visible at boot time, account password[2], etc.) in an email or voicemail to the TSA.
2. Once you receive confirmation that your item(s) have been found (along with some sort of property tracking numbers), create a FedEx account with a nine-digit account number. The TSA requests that you verify that this account number can be used to purchase shipping supplies, but I really couldn’t get anything but useless form responses from FedEx about this — and then it worked when they did it.
3. Send the FedEx account number along with a shipping address to the TSA, including the property numbers you received and some shipping instructions (such as whether to send overnight, three day, etc.).
4. Wait and pray.

Creating a FedEx account from overseas and configuring payment methods was a real pain; I assume this is because they assume fraud, which I appreciate, but still. There appeared to be no way to prove legitimacy, setting up payment information simply failed with no explanation. The reason I believe overseas connectivity was the problem is that I was able to set up an account using an SSH tunnel back to a VPS in the US as a proxy. (If you don’t know about ssh -D <port>, look it up now.) With that in hand, I completed the list above. The TSA never contacted me again to let me know that my laptops were shipped, or even to confirm that both items had been identified. I had reason to believe they had shipped only because I received notification from FedEx that an item had been shipped on my account.

If we fast forward a couple of weeks, we get to receipt of the laptops. When I received them, they were ludicrously poorly packed. The two laptops were individually wrapped (reasonably well, fortunately, for this is the only thing they did right) in bubble wrap, and then very loosely (by very loosely, I mean such that they could be separated by about 12") bound together with bubble wrap and packing tape ... and dropped in a box that could have held literally six or eight such bundles, and enclosed a volume of ten or more.

I’m not sure I’ve ever seen a more irresponsible or poorer packing job, and certainly not from anyone who should be handling their packing professionally. If there’s any clearer sign of the incompetence and general lack of professionalism at the TSA, I don’t know what it might be — unless it’s reducing elderly, minor, or infirm passengers to tears with invasive and demeaning “security” procedures, of course.

Now that the story of what happened and how it was mishandled is out of the way, let’s move on to what can be learned from this. Some of it is obvious, and even things I already knew (or should have known).

• Protect your data. This seems obvious, but I know that many people don’t follow it. Both the operating system and my data are encrypted on my laptop. Unfortunately, the entire chain of bootup can’t be completely trusted, because a) I have no way to trust the BIOS, and b) the boot partition must be unencrypted (though, once booted, preferably from external media, I can verify that it is unchanged via my package manager), but at least I can have some reasonable assurance that my data was not stolen.
• Mark your property, and retain serial numbers. I was able to get my laptop back only because I just happened to have one of the uniquely identifying items on its serial number plate. I don’t have that for most of my other electronics. I plan to get property tags of some sort for large and expensive items. I know several people who do so, and it now seems prudent. I keep serial numbers for all sorts of things for insurance purposes, but it honestly hadn’t occurred to me to keep them for travel purposes. I don’t know why not.
• Carry duplicates of important data in separate luggage. I had carefully copied a fair amount of data that I knew I was going to need abroad to my laptop before departing ... which did me zero good when it was left behind in the US. I should have had an extra copy on a USB drive or similar medium, possibly checked in my luggage (and encrypted, of course). As it is, I wasted a fair amount of time dragging several hundred MB or more of data across the Atlantic and through a poor cellular connection.
• Bring a second computing environment with you. I don’t mean another laptop (that would have been lost, too), but something like a live USB disk. When you get where you’re going, it’s entirely possible that you’re going to find yourself in a primitive and unhospitable land with no access to anything but (for example) Windows. I spent another large chunk of time creating an environment in which I could work without reinstalling a borrowed machine. Checking a cheap (and therefore reasonable to subject to the vagaries of airline baggage handlers) machine such as a Chromebook or older netbook might be a reasonable solution, as well. For my own part, I plan to customize an Ubuntu live USB and put it in a different carry-on from my laptop.

So when all is said and done, I have my computer back, my data is intact and safe, and all I’m out is a few hours of backing up and restoring. I was due for release upgrade, anyway. One could claim that the above criticisms are first world whining, but I strongly disagree. If my laptop had been randomly lost in an airport somewhere because I set it down, then I would consider the successful recovery a lucky victory. The fact that the only reason it was out of my bag was for flawed and ineffective security theater, and that the only reason I left it behind was that I was pulled aside for additional theater, changes the equation. Instead of an unhappy accident with a happy ending, we arrive at an unacceptable travel interruption and onerous burden of loss with a bungled response.

I have a Baratza Maestro Plus high quality consumer-grade coffee grinder for press/drip/pourover/etc. preparation that, as best I can figure, I've had for somewhere between seven and ten years. A couple of weeks ago it suffered a catastrophic failure due to violence not of its own making. Much to my surprise, not only was it readily repairable, but Baratza appears to go out of their way to make this a simple and pleasant activity. That sort of attitude is sufficiently rare these days that I thought it was worth a mention.

A Quick Review

The Baratza Maestro Plus is a now-discontinued conical burr coffee grinder targeted at enthusiasts who primarily brew coffee via one of the methods typically employing a coarser grind, such as pourover or press. When it was released it cost somewhere just under the $200 point (depending), and represented one of Baratza's more entry-level offerings. As mentioned, it is now discontinued and has been replaced with either the Encore or the Virtuoso, depending on your viewpoint and preferences. (In my opinion, it fell somewhere between these grinders, and compares more fairly with the Virtuoso, which is targeted at a somewhat higher price point.) Within its grind range it has a fairly flexible selection of grind size, but it doesn't get quite fine enough for espresso or Turkish with the kind of precision and grind consistency those methods ask for. I've used it for Turkish when I didn't wish to incur the tedium of a mortar and pestle, and it answers, but it's certainly not ideal.

It has a number of touches that cement its enthusiast-class reputation, such as a nice heavy cast metal weight in the base to give it stability (particularly for using the momentary push button grind trigger) and reasonably consistent burrs that produce a grind that, while it has some size variation, is a wide cut above the typical department store burr grinders. It also has some annoying misfeatures, such as a timer knob that falls off of its shaft with regularity. (A touch of silicone cement seems to have cured that particular ill for me, while still allowing the knob to be removed.) It is reasonably easy to clean and looks decent on the counter. The quantity of ground coffee trapped in the ejection chute between grinds is fairly moderate, which I understand can be a failing on some grinders in this class.

The failure

My particular Maestro Plus failed through no fault of its own — a small pebble managed to make its way through winnowing, roasting, and winnowing, and then into the grinder for my morning brew. (I blame this on roasting first thing in the morning, when I'm not at my most alert!) It wedged itself thoroughly between the stationary and moving burrs, and (as I found out when I disassembled the grinder) caused a metal pinion on the motor shaft to shear the teeth off a small section of the plastic reduction gear on the burr shaft. When I say it was thoroughly wedged, I mean that it was thoroughly wedged. I had to use more force than I was comfortable with to get it broken up so that I could pry it out from between the burrs for disassembly.

The immediate symptom of the failure was that the moving burr did not rotate reliably, and when it did, just a few beans between the burrs would stop it (by catching it when the stripped portion of the reduction gear was straddling the drive pinion). When it did rotate, it made a horrible periodic clicking/crunching noise.

The fix, and excellent support

Upon this failure, I said to myself, “this grinder is pretty old, let me see what's on the market.” I rapidly found out that what is on the market is grinders that cost more than I want to spend for dubious improvements (or not) over the Maestro Plus. I didn't want to spend the money to really trade up, and it looked like the closest to a sideways trade I could accomplish would have been the Virtuoso at somewhere north of two hundred dollars.

Being a fix-it-up kind of person, I decided at this point to see what was wrong with the unit and whether I could coax it into working. I quickly located the aforementioned pebble and removed it, but found that the grinder was still not reliable (due to the stripped gear that I had not yet seen). It was not immediately obvious how to get the cover off without damaging the plastic, which was disappointing to me — but when I did some searching I rapidly found that Baratza actually has a detailed PDF explaining how to remove it! This was surprising and gratifying.

I removed the case and immediately found the stripped gear, so I headed back to Baratza and looked to see if I could find parts ... which I immediately found on their parts support page, another gratifying happening! Not only do they have an extensive list of parts for both current and discontinued models, but they have upgrade parts and very reasonable prices. I ordered a gearbox rebuild and upgrade kit as well as a stationary burr holder (which had been cracked for some time, but never failed).

I want to take a moment here and throw in an aside, which is that the engineering of this grinder is really excellent where it matters. The motor, gear train, rotating, and stationary burr are all mounted together to a stiff plate (metal in the original version I had, and some sort of filled plastic with stiffening ridges in the rebuild) that forms the working part of the grinder and is really just mounted in the case for convenience and stability on your countertop. This entire assembly is rubber mounted in the chassis via little brass-and-rubber bushing assemblies. The chassis also contains the controller board (which I did not examine closely, but appeared to be mostly passives and a discrete bridge rectifier, possibly just a DC power supply), a safety lockout micro switch that prevents the grinder from operating when the hopper is not installed, and a mechanical timer and the momentary trigger. Wiring between the various components uses push-on blades. I see no reason this design shouldn't last forever.

In just a few days (I think I ordered on a Sunday and received the parts on a Thursday; I opted not to pay for express shipping) I received the replacement parts for a grand total of about $22. Using some really excellent instructional PDFs from the Baratza site, I was able to quickly install the redesigned gear assembly (critically containing a replacement reduction gear!). At this point I noticed that the rebuild kit was missing a couple of M5 Allen head cap screws that the instructions referenced. I called up Baratza to find out exactly what they were supposed to be, and in just a few minutes I had the answer: M5x12. The representative I spoke to offered to send out replacement screws immediately, but I opted to spend $0.66 at the local True Value Hardware instead, just to get back in operation faster. I don't consider this a major failure on the part of Baratza, because a) the missing parts were cheap, b) they assisted me in locating replacements, and c) they offered to make it right by sending out the missing parts immediately and without question.

The reassembled grinder works admirably. The grind sizes are a bit different on the numeric scale, so it took me a few tries to dial in a grind, but the consistency and quality of grind are good. The new gear train is quite a bit louder than the old gear train, but I seem to remember that the old gear train was louder when it was new; perhaps this one will wear in a bit and quiet down.

Closing remarks

All in all, I am very pleased with Baratza for producing a fine, well-engineered product, and then standing behind it. Very few companies in the 21st century can be bothered to provide parts and repair information to the consumer, much less via the support page on their web site with a simple shopping cart and order process. Baratza's attention to customer service kept a serviceable grinder out of the landfill for want of a $0.50 gear at a price that is perfectly acceptable to me as a consumer (and only about 10% of the price of a new grinder!) after nearly a decade of service. I look forward to the next decade with this grinder — maybe I'll put new burrs in it before then!

I've been using X11 for a couple of decades now, through several brandings and incarnations of the X server, but it's a sufficiently huge and complicated stack that there's always room to learn something new. Today I had an eye-straining contrast problem with an Intel i915 graphics chipset, and learned a bit about the xrandr tool that I hadn't known before — it can control color profiles and configuration.

TL;DR: The fix is here.

I just installed an IOGEAR HDMI KVM Switch on my primary workstation, because I have a whole pile of PCs and PC-like devices sitting here that I almost never need a console for, but when I do I don't want to have to dig all around under the desk to get them hooked up. This necessitated moving my primary workstation's video output from DVI to HDMI; since it had a built-in HDMI port, I didn't bother to buy a DVI-to-HDMI adapter for it, I just plugged it in and went with it. When I first got everything set up, I didn't notice any particular effects, and I used it for a while. After some time, I started to notice that the display was too bright and washed out, and that it was giving me eye fatigue. It looked like the backlight was set too high.

Uh oh. When you spend a lot of time in front of a computer, decent display quality is an absolute must. This situation wasn't going to be acceptable. The first thing I did is hook up the old DVI-D connection to make sure that it really was better — and it was. The second thing to do was ...

I really had no idea. I had some vague idea that xrandr might be able to let me set the backlight, but I'd never used it for that, and the manual page said that the xrander --backlight option was a software-only change, and wouldn't affect the actual brightness of the monitor. It suggested xbacklight, which it turns out does not work on this chipset/monitor/something.

So I did what any computer user worth his or her salt would do — I asked Google. Which had no idea. (In point of fact, I asked DuckDuckGo first, and when it didn't know, then I asked Google, which didn't know either.) This caused me to start to set about determining whether the problem was the HDMI output itself, or something to do with the KVM. A quick comparison of the four devices currently hooked to the KVM input showed that:

• My development workstation's DVI output was much better than its HDMI output.
• The Linux box on the second port didn't seem as washed out at the console, but it didn't have working X11. The backlight seemed tamer than port 1 or port 4 (coming up).
• The Mac Mini on the third port seemed ... great. It was hard to tell, because I had a difficult time finding a window with high contrast and a dark background that wasn't faded, semi-transparent, dancing, or otherwise behaving stupidly.
• The Raspberry Pi on the fourth port looked terrible. This shouldn't surprise anybody; they're really designed to hook to a TV, for crying out loud.

Being a somewhat inconclusive test, I decided that installing an X server and my basic desktop config (I use FVWM as my window manager and rxvt-unicode as my terminal, so blasting a minimal configuration to a new machine is copying one directory and one file) on the second Linux box was going to be a good comparison, since its console looked better.

To make an already long story short(er), it turns out that the second machine's Intel Haswell-based graphics chipset wasn't supported in Debian Wheezy. This led to more Googling, which turned up this gem (which is unrelated to the Wheezy-Haswell problem, but solved my original problem):

xrandr --output HDMI2 --set "Broadcast RGB" "Full"

This magical command fixed the problem. The HDMI output looked just like the DVI output! The i915 appears to support three values for the Broadcast RGB parameter: "Automatic", "Full", and "Limited 16:2". Automatic for the DVI output defaults to Full, while automatic for the HDMI output defaults to Limited 16:2. As best I understand it, this is because many TV panels don't actually support a full color space, so the graphics card can squash the colors into a color space supported by those limited panels. On the HDMI output it defaults to doing so, so that if you hook a TV up to the HDMI port the colors will be usable. What had appeared to be a backlight problem was really light leaking through not-quite-black pixels.

I learn something new every day.

Credit cards play an interesting role in the security of our finances — or they can, if used well. They insulate our money from the institutions with which we do commerce, by placing the issuer in between. The issuer typically makes relatively strong guarantees about our liability in the event of misuse (often, in practice, assuming the entire cost of a stolen card number in terms of real dollars, if not time and effort), in return for shaving a few percent off every purchase when the card is used. Nonetheless, they are a handle into an individual's finances, however indirect, and recently every party in the chain from consumer to product has proven somewhat irresponsible with them. Consumers rack up charges they cannot afford; retailers, processors, and issuers lose transaction data and card numbers to employees and external criminals; issuers extend credit in ludicrous amounts, etc.

Given the general environment of irresponsibility with credit card data (Target, Neiman Marcus, and Home Depot being three retailers with recent very large scale breaches, affecting a hundred million or more total consumers), it appears that at least some retailers are taking steps. Today I tried to return some merchandise to a JC Penney home store with the original receipt, and was told that I could only take in-store credit unless the credit card used on the purchase was physically present. When I inquired as to why (since I haven't experienced this with a receipt in hand in many years), the teller said that JC Penney no longer stores credit card information after the transaction! (Presumably they keep it for some time until the transaction is actually cleared, and then purge it, but he was not specific about this.) I was both surprised and gratified to hear this, and it somewhat made up for the annoyance of being unable to return the items (since they were purchased on my wife's card). They could, of course, have returned my purchase price in cash, but I do understand that they don't wish to lose the card spread and fees in that manner.

I hope this is true, I think it's a good move if it is, and I hope it represents the start of a trend. It may be convenient to be able to walk into a store without a receipt and return an item based on credit card history, but it's not good security. I've certainly enjoyed that convenience in the past. I would give it up without complaint if it meant I didn't have to do the new-card-dance every couple of years due to a widespread data breach.