Monthly Archives: May 2014

What do Americans and Finns know about Ammonia?

Via Jon Eisen on twitter, Hal Levin at MicroBEnet has gotten himself into trouble with this post (now deleted as not reflecting the views of MicroBEnet). The part that annoyed readers (including me, after seeing the tweet calling it out).

On May 8th, CNN reported on “Top 20 most polluted cities in the world.” An esteemed colleague sent an email today to a small, informal discussion group of building scientists (the majority are former EPA scientists), that read as follows:

“Only in America would people be dumb enough to talk about the “diameter” of a non-spherical particle. I bet even in Finland they know that ammonia is a gas and not a particle, and I’m not aware that these others mentioned are carcinogens. .
From the web today:
CNN reports, “PM2.5 refers to the diameter measured in microns of particulates such as ammonia, carbon, nitrates and sulfate — which are small enough to pass into the bloodstream and cause diseases such as emphysema and cancer.”

I hesitate to provide the link to the CNN post, but if you still want to, you can find the CNN story at

If you follow the link to the tweet, you can see that I weighed in with some replies. As an alternative way to procrastinate the work I should actually be doing on a Sunday afternoon, I decided to dissect this. The first irony about “only in America” is that CNN reporter Madison Park is currently working out of the Hong Kong Bureau of CNN. I suspect she could be an ABK, and she went to UC-Berkeley so I guess we can blame America for her reporting. But she was a journalism and history major, so despite covering health for CNN, I suspect she defers to her sources in writing things like the passage that offends Levin’s email correspondent. The information she’s reporting on is from the World Health Organization and searching their website for “ammonia air pollution” gives this page from the WHO media centre as one of the hits. What does it say about ammonia and particles?

Particulate matter

Definition and principal sources

PM affects more people than any other pollutant. The major components of PM are sulfate, nitrates, ammonia, sodium chloride, black carbon, mineral dust and water. It consists of a complex mixture of solid and liquid particles of organic and inorganic substances suspended in the air. The most health-damaging particles are those with a diameter of 10 microns or less, (≤ PM10), which can penetrate and lodge deep inside the lungs. Chronic exposure to particles contributes to the risk of developing cardiovascular and respiratory diseases, as well as of lung cancer.

There must be some way to blame George Bush or the Koch brothers for this info on the WHO website, right?

It’s somewhat interesting that an email list full of former  EPA scientists might have let this pass, as I would imagine that some of them might have worked with their counterparts at WHO. But not being privy to this email list, I can’t assume that nobody pushed back.

I was also struck by a couple of other things:

Only in America would people be dumb enough to talk about the “diameter” of a non-spherical particle.

There are plenty of things that have diameters that are not spherical. Circles, cylinders, and cones, for example. We speak of the diameter of the earth although it is not perfectly spherical. Moreover, even highly irregular solids, such as proteins, can be discussed in terms of their effective radii and diameters: the Radius of gyration. From the point of view of passing through filters or blood vessel walls, this seems perfectly appropriate to me.

I bet even in Finland they know that ammonia is a gas and not a particle

In the twitter discussion I wondered what the emailer has against Finland. But here I wanted to discuss a few other things:

  • As long as we’re being nitpicky: Ammonia is a gas at ambient temperature and pressure in the cities being reported on, but it’s not always a gas.
  • The average American thinks of ammonia as a liquid because they can go to the supermarket and buy cleaning products labeled as ammonia. The one shown is actually from Australia, so even if they don’t have this kind of stuff in Finland, it’s unlikely that usage that fails to distinguish ammonia from ammonium hydroxide solutions is limited to the US.
  • Unless I missed a revolution in chemistry, gas molecules are particles.

The decision to take it down at MicroBEnet makes sense, insofar as the goal of that site is to educate the public about the microbiology of the built environment, not to divide readers into those who agree and those who are offended. Insofar as is funding a collaboration between Eisen and Levin, this kerfuffle may have some unfortunate hangover effects for an interesting project. Let’s hope this doesn’t poison the built environment.

What I find both sad and interesting in the now-deleted post is: What drives someone to take a set of minor issues in a news story that are arguably not even errors and not only decide that this reflects a kind of egregious ignorance that is unique to one country, but also share that opinion with a bunch of strangers? It seems likely that both the emailer and Levin thought that this view of Americans (and Finns?) is not only accurate but also obvious.


I’ve been frustrated by Apple’s Podcasts app on my iPhone, so I decided to try InstaCast on the recommendation of a friend on twitter.  Initial impressions:

  • The first thing I noticed was that when you launch it for the first time, it doesn’t ask if you want to import Podcast subscriptions from iTunes or the Podcasts app.  There is a way to do this but it only imports the subscription URLs, not which episodes have already been played or not.
  • The app has a new UI optimized for iOS7… which means that the online docs don’t seem to show screenshots that match the app
  • Some podcasts only show the most recent episode. Not clear if I can get the old ones
  • I hate having podcasts in the cloud. This is one of my two main frustrations with the PodCasts app (the other is it deferring to the Music App). I do not want my data plan sucked dry pulling podcasts over the cellular network when part of the reason I got the 32G iPhone in the first place is so that I could download podcasts and listen to them when out of range of wifi. InstaCast can download episodes, but the UI to download them is a bit cryptic. Comparephotophoto
    Clicking on the cloud icon in Podcasts seems intuitive by comparison. In InstaCast, you have to click on the menu in the bottom right, select podcasts, and then click the download icon that appears in the bottom bar after you’ve selected the menu icon. What is worse it that the top item that shows as ready to play has not actually been downloaded yet!
  • If you want to see which podcasts have actually been downloaded, I think you have to go to Lists > Downloaded… which you can’t filter by source. You can’t tell by selecting an item in the subscription list and looking at the download icon. It doesn’t gray out even if that podcast was already downloaded.
  • When you click on the menu icon, you get a Select All option. But as soon as you select one, the option to select all goes away. It doesn’t change to Unselect All. So if you select half of a list and then change your mind, you have to cancel and then come back to Select All.

So far, I’m not impressed.

RoomScan Pro

Today we went to a seminar about remodeling with Stearns design-build to get some ideas about some remodeling plans. In talking about things, I decided that we really need a better diagram of our house layout. It doesn’t have to be super precise, but I want a starting point to sketch from. I thought there might be an app for doing measurements and it turns out there are a couple

  • Magicplan uses the camera. It sounds interesting but this review makes me leery of the “subscription” part. It also seemed to me that there could be issues with obstructions.
  • RoomScan from locometric works using the GPS, gyroscope, the camera and who knows what else (accelerometer?) to take measurements based on tapping your phone against the walls

As with anything that involves extrapolation from the phone to do a complex task, getting dimensions right with RoomScan isn’t quite as easy as it looks in their demo video. But it’s still pretty cool, and their support people are very fast to answer crossword questions via twitter to @locometric.

Testing – technique matters


Figure 1

In my first attempt to create a floor plan (Figure 1), the numbers were off. The app did warn me about the measurements possibly being off. RoomScan does provide a way to adjust the lengths of walls manually, but some of these were way off. Could this be due to the extended wifi network? Are the numbers reproducible? Was I doing it wrong?

RoomRoomScan L (EW)RoomScan W (NS)L W
Master BR11.511.513.7514
Master Bath10.5 in 2 segments119'1"7.5
Main Hall263303.5

I decided to do some experiments. First, I repeated the hallway a couple of times. After two failures, I switched the phone to Airplane mode. I also changed the tapping pattern (see below). RoomScan worked fine in Airplane mode, showing that the wifi and cellular signals aren’t being used. Once I got it right, I switched Airplane mode off, and did more replicate scans. Figure 2 shows the last test.

Test 1Failed
Test 2Failed
Test 3313.5
missed a door
Test 430.5
21 to BR door
Test 530.5
21 to BR door
Figure 2

Figure 2

So, I think that the problem with dimensions is that I wasn’t moving fast enough for the accelerometer to work optimally. RoomScan warns you about this. There’s a voice message that says things like “I work best if you move quickly but smoothly” or “Keep the time between taps short” (not transcribing these exactly”. It appears that RoomScan uses the lack of light from the camera to determine that you’ve tapped a wall.

It also took me a couple of tries to figure out what it wants to close the polygon. It’s not enough to tap on the same wall past the starting tap. For these, I started on the lower wall to the left of the BR door and went counterclockwise. At the opening I did wall tap, opening tap, opening tap on the opposite wall, wall tap before heading down the long part of the hallway. Coming back, I had to go through through that whole sequence a second time to close the loop.

Adding rooms

Next, I deleted all the hallway tests but the last one, which I used as the hall. Unfortunately RoomScan wouldn’t let me rename it. I got the Pro version because it wasn’t very expensive ($4.99) and it allows you to add rooms through doors. You open a previously scanned room and touch a door or opening as a starting point. RoomScan brings up a control wheel with different options. These include the ability to change the kind doors to openings and vice versa, and flip which way the doors open and which side the hinges are on. Tapping a wall gives a similar control wheel that allows you to add doors that you didn’t scan in earlier.

RoomScan control wheel

Figure 3. Control wheels for editing adding rooms

Selecting add room through door/opening prompts you to name a new room and then click Add. Depending on what kind of connection you start at, it tells you to start in the middle of a closed door, or on one side of an opening.

In my initial tests, I didn’t get the opening at the end of our hallway right, so I used the add door function to create an opening.  Starting with an opening you’ve added via the edit function doesn’t work right. I’ve emailed service@locometric with screenshots, and they replied very quickly (on a Saturday no less) promising a bug fix in the near future regarding the openings behavior. But the add room through opening works fine if you create the opening in the original scan.

Figure 4

Figure 4. tapping to get an opening.

Figure 4 is an image RoomScan support sent me to show how to do it right. Essentially, you double-tap each side the opening: once to set a wall point and the second time to set an opening point. The order goes wall-opening-opening-wall.

Once my scanning technique got better, I was able to add several rooms off the Test 5 hallway. Figure 5 is what I have so far. RoomScan automatically places each room based on the opening or door you previously started with. You can adjust this by tapping a room in the overall overview and dragging it to a new position. My three problems with this are

  • It isn’t precise enough. It would be really nice to be able to just nudge a room by a little bit. You can see the nudge problems in how the rooms line up.
  • Roomscan’s floor plan view tries to merge openings and doors into nearby spaces a little too aggressively.
  • It tends to crash in this mode

I want the connection with the opening I started with, and there are times when the openings do need to be connected. For example, our dining room has openings into both the living room that should be connected when I scanned the Dining Room starting from the Kitchen with the Living Room already on the floor plan. This connection is presumably made by recognizing that two openings from adjacent rooms are overlapping. But the overview also tried to fuse two kitchen closets into the back of the Master Bathroom. It didn’t create new doors in the bathroom, fortunately, but it was disturbing to watch the doors connect like cytoplasmic bridges between cells as I tried to move the Kitchen into place.

I also had to do some fudging on where I think the boundaries of the rooms are. The way our flooring is set up, I think of the boundary between the living room and Kitchen as being diagonal. I don’t think RoomScan likes diagonals. I’d be interested to see how it works on floor plans with different angles between the rooms.


Figure 6. The partial floor plan


RoomScan Pro is not going to replace a professional draftsman spending days and $$ to generate new architectural plans for your house. It’s not going to give you elevations. But at $4.99 it’s a lot better than what we were doing for early stage brainstorming – running around with a too short tape measure and sketching on paper. That said, what I think I’ll end up doing is:

  • Tweak the measurements in the app and export the image
  • Import the image into Notability or a drawing app and sketch on top of it on my iPad.

That’s partly because I’m not sure how to export RoomScan output as objects to a drawing program where they can be edited on my laptop (It’s not clear that any of the home/interior design apps handle that well either; what are the file format standards, anyway). It’s also because even if I could export editable files, I don’t currently have anything to import and edit them.


Who gets NIH grants and is it a problem?



Drugmonkey posts a graph from showing distribution of grants by institution. A series of follow-up posts at Grantome includes one that looks beyond the top 50 institutions and concludes:

The NIH website reports that more than 80% of its budget goes to over 2,500 universities and research institutions. Yet for the R01 – which is arguably the most sought-after, prestigious grant that is the financial staple of many medically oriented research laboratories in the U.S. – the bulk of the grants are distributed to 4% of these 2,500 institutions. It therefore appears that the division of R01 grants among institutions parallels that of wealth among individuals in the U.S. population. It is highly unequal, with the bulk of the pool held by a lucky few.

Drugmonkey asks both in the comments and on twitter:

50 institutions get 60% of the #NIHGrants. Good, bad, or meh?

The distribution of grants is interesting, but it leads to the obvious question: what should it look like? I don’t think that is knowable from the data provided, and it may not be knowable, period. The description of the top group as “a lucky few” suggests a perspective from a particular point of view. As with the reactions to the Alberts PNAS paper (which is now taking comments btw) scientists/PIs tend to view the question from the perspective of what is good or “fair” to scientists. But  it is important to remember that just as pro-market is not the same as pro-business, what is good for Science is not always the same as what is best for scientists. Or, as I tell graduate students about the importance of working hard: the NIH is not supposed to be a welfare program for smart people. I usually add what Bob Sauer used to say to every new person in the lab: you don’t have to be that smart to do Biology. If you were really smart you’d be doing math or philosophy or shit like that.

The correct but painful question then is not whether PIs are being treated fairly, but rather whether federally funded science agencies are doing a good job of optimizing the taxpayer/citizen return on investment in research. In the case of NIH, the investment is in biomedical research (preferably a big tent definition of biomedical, IMO). In general, while I still tend to think that the peer review systems at NIH and NSF are like democracy (terrible but better than most the plausible alternatives), I do worry that the system is getting worse at approximating an ideal system for maximizing the taxpayer investment in research. But I am not sure whether a system with the nonexistent perfect program officers , SRAs, and reviewers would increase inequality or decrease it.

Lots of interesting reading at the links, in any case.

Do architects get what we do in the lab?

This post is what I wrote in February of 2009 on the old blog. I was prompted to dig this post up by Virginia Postrel tweeting about this article in Slate. I pulled the text from a database dump of the old blog. The images are old too, but were not in the original post (found them in my iPhoto library).


Derek Lowe discusses the new biochemistry building at Oxford (see Nature and their VR tour). Derek:

… I kept wondering, where have I heard descriptions like this before? Oh yeah, the last time I moved into a new building. Actually, every single time I’ve moved into one, come to think of it. I was part of a gigantic corporate move in 1992 into what was billed as a “high-interaction facility”, which was nothing of the sort. And then at the Wonder Drug Factory, one of the new lab buildings had the whole research area behind a large glass wall; it was the first thing you saw when you came into the place. Unfortunately, since it was full of snazzy equipment, it became part of the standard tour for visitors (the combichem labs were largely abandoned by then), and the people working there sometimes felt like zoo animals. And my current building has the labs all around the outside walls, and a huge atrium in the middle of the building (to what purpose, no one is sure; it’s completely empty).

Microbial Sciences Building at Wisconsin. The “high-interaction” propaganda reminds me of the junket I was on a few years ago where we visited Stanford, UCSF, and Chiron. The Clark Center at Stanford was an extreme case of the open floor plan concept to promote interaction, and as far as I could tell, that was a failure. As Derek’s commenter Kako notes about (perhaps) a different building, a big problem with open floor plans for labs is that they are really noisy. What we saw was lots of people with iPod headsets, small equipment rooms being taken over to be used as offices, and the portable furniture being rearranged so people could get a small amount of privacy. We heard that the original plan was to have the PI offices in the midst of the open plan… until someone pointed out that this was not going to work when the tearful undergrads needed to see the profs about their grades. Also, Kako points out:


Atrium of the Microbial Sciences building at Wisconsin

Security – every door is open. Bye laptop! Bye purse! Gradually getting better, if only cos people learn *never* to leave things unattended.

The Clark Center did have something that was a successful high-interaction space: a Peets Coffee shop on the top floor. I spent many afternoon breaks there when I was on sabbatical.

The Genentech building at UCSF suffers from overblown scale, but the arrangement of the labs and offices struck me as smart. The faculty officesThe comment about the atrium reminds me of what some of my friends said about the clustered around common spaces, so the PIs and members of different labs would interact in each “pod” of groups. People interact in those spaces, but can retreat to a more private space to focus in smaller groups, or alone.


A lab in the Clark Center. Note the equipment room taken over as an office to get some privacy

Looking at the virtual tour views of the Oxford labs, it looks nice, but I’m wondering about the way the desks open onto the shared space. I suspect there will be a lot of iPod headsets used there too.

When I first got to TAMU, I had more interactions with people because my office was close to the rest rooms on my floor. Debby has argued that interactions can be promoted by putting restrooms and the mail room on opposite ends of the building. Not sure if that still works anymore since so much correspondence is by email. In BioBio, having lots of common use equipment promotes interaction. Having more whiteboards in the halls would promote interaction – I’m not seeing those in the Oxford pics.

Adventures with the iPhone music app

My wife and I like to joke that stores do their market research by figuring out what we like a lot and then discontinuing those items. That’s how I feel about Apple’s decision a couple of iOS upgrades ago to decouple Podcasts and iTunes U from Music.

I like to listen to podcasts. I listen to podcasts much more often than I listen to actual music. I often listen to podcasts using my iPhone hooked to the USB connector in my Honda Civic hybrid. Before Apple separated the apps, my phone would play the next podcast when it finished the episode I was listening to. Because I have a very short commute, I usually have a bunch of episodes to catch up on.

After the apps were separated, my phone tended to default to the music app instead of continuing to play podcasts. This was annoying but I tried to avoid the safety hazard of fumbling with my phone while driving (at least until a stoplight). It wasn’t so bad… the song that came up was usually Miles Davis’ All Blues, one of the greatest jazz standards of all time.

But that was before the most recent iOS update, which seemed to have turned on some kind of cloud sync with my iTunes account, even though I had it set to “Sync selected playlists, artists, albums and genres”. A couple of years ago I downloaded the free Apple Christmas music compendium, which included the incredibly insipid and annoying Above the Northern Lights by Mannheim Steamroller… and until just now I couldn’t delete it.

  • I deleted it from my laptop and resynced. It was still there
  • I swiped left and got the delete button on my phone. I hit delete. It came back
  • I tried the above with many different permutations of settings on the phone and the iTunes app on my laptop. It was still there.

Many of these attempts were based on Finally, I found this thread on the Apple forums. The solution involves deleting all of your music and then resyncing with a computer.

  • On the phone:
    • Settings > Music: make sure Show All Music is Off
    • Settings > General > Usage > Music: use the edit button to delete all the data from the music app
  • Connect to the computer and resync to copy music from iTunes to the phone

Note that this doesn’t work in the untethered world that Apple aspires to for iOS devices. Also, when I turn Show All Music back on, the damned song comes back. I never subscribed to iTunes match, btw.

As an Apple stockholder, I usually just let my proxy default to the current directors. But I would consider voting for an alternative board whose platform was to fire whoever at Apple is responsible for splitting podcasts out of Music (deleting the annoying song was not a priority for me until the stupid cloud song became the default) and whoever made it so freaking hard to delete songs.

Human subjects and education research

At Retraction Watch, there’s a story of a paper about ethics training retracted due to IRB human subjects protocol problems. We tend to think of human subjects research as involving things like drug trials, but a lot of it is things like this:

This was an IRB-approved paper-pencil study investigating how certain features of ethics case studies influence knowledge and application of case study principles to new ethical scenarios.

In other words, if I’m understanding the post and the excerpts, the investigators were studying what works and what doesn’t work for how to teach higher levels in Bloom’s taxonomy in the field of teaching about ethics. So what went wrong?

Several administrative issues influenced our Institutional Review Board’s decision to not allow the data from this study to be used for research purposes. One of these had to do with the fact that some of the course instructors were not listed as key study personnel and they handed out and collected the study materials and informed consent forms. Even though they did not have any other involvement in the study, we recognized this oversight. Additionally, we implemented two minor changes to study materials, including dropping two items and renumbering 8 items, and did not obtain re-approval for these changes. Lastly, through this review process, we became aware that roughly half of the informed consent forms (ICFs) were not on file. Although we kept a clear record of who consented and who did not through the use of a training checklist, we recognized this was a data storage lapse. We worked with our IRB to fix these problems and have better processes in place to prevent similar issues from occurring in the future. Although the senior editor for the journal did not think that these issues warranted retraction of our paper, our university’s decision that we could not share the data publicly influenced our decision to voluntarily retract the article.

I understand the basis for the IRB/Human subjects rules, but this case illustrates a problem for all of us as educators. Every time I teach any course, I am doing an experiment in what works. It’s usually an uncontrolled experiment with unconscionably bad record-keeping, but it’s still an experiment and my students are subjects. As there is an increased emphasis on doing more rigorous assessment of student learning outcomes, we will try to be more rigorous about gathering and analyzing data from assessment instruments, which can be done without IRB approval as long as it’s not for publication. Consider how different that is from a drug trial!

As I understand it, doing a proper human subjects protocol for assessment of teaching requires outlining ahead of time all the interactions the assessor will have with the students/subjects. Deviation from that protocol (analogous to changing an FDA approved drug manufacturing protocol) is a no-no – as in the case above. This makes it almost impossible for the actual instructors to also do the assessment… and creates a barrier to dissemination of interesting STEM teaching methods being tried by scientists who don’t have the right collaborators for assessment.