Google goes stupid

Yesterday, at work, I was using Google Image Search to try to find an example of a particular kind of scientific figure. I wanted a figure where any variable has been measured over ages from early in life to late in life, and where the data are U shaped (start out high, go down, then go back up) but where someone had fit a simple linear regression to these data, completely missing the pattern. I see such figures fairly often, but couldn’t remember exactly where. I also wanted examples where they had got it right and fit something other than a straight line.

So I used search terms like: “linear regression” “scatter plot” age and figure. And I got back images of babies, mothers breastfeeding, pornographic images of women (not medical or educational or artistic, despite having Safe Search set at strict) mixed with a smattering of scatter plots, almost all from papers about women’s reproductive health, breast- and cervical-cancer and baby nutrition. None of it was useful to me, and a lot had no apparent link to any of my search terms. What was going on?

At the same time at home, my wife was reading a Blogger hosted blog on women’s reproductive health and baby-care. She was using my laptop and was still signed into gmail as me. Blogger is owned by Google. Yesterday was the first day of Google’s new policy of sharing information about users of any of its services with any of its other services to help to personalize search results and the ads it shows. I was looking for figures. My wife was reading about women, their reproductive health, and their babies. I got images of babies, women’s figures and reproductive parts, and scientific figures plotting data on same. If I was interested in scatter plots related to breast-cancer, I would include “breast-cancer” in my search terms. Yesterday, I logged out of my Google account in order to get decent results from Google. This is not helpful for Google or for me.

There has been a big kerfuffle about the lack-of-privacy implications of this new policy. People say that Google is becoming Facebook, a sinister ploy to know more about you than you do. I am sympathetic to those complaints, but not particularly worried about them for my own sake. My main complaint isn’t that the new policy is evil (i.e., Facebookish), but that it is stupid (i.e., Microsoftian), in that it makes Google’s core product, Search, less useful. Google is forgetting what Microsoft has never figured out: Too many functions doing things for you detracts tremendously from the functionality of the software. If I want to sit down at a new computer and use Word, I don’t want to first have to individually turn off all twelve parts of Autocorrect and 75 other things that will change my document for me in undesirable ways. And if I want to search for something, I don’t want terms I haven’t inserted invisibly added to my search. To have the algorithm decide to include terms from a blog that I (or anyone logged into the same account) read in my search takes control away from me, and then I have to fight against the algorithm to find what I want (or, heaven forefend, use Bing). It’s like having a car that tries to drives you to a restaurant of its choice every time anyone in the car, or on the radio, mentions food. You drive because you have to, but after the tenth visit to that terrible restaurant, you wish you could just walk.

The Google–becoming-Facebook complaints are surely in part a reference to Google’s relatively new social network site, Google+. But here again, it feels to me more like Google-becoming-Microsoft. My complaint stems from their push to drive every possible bit of traffic to Google+, even if this makes their other products harder to use. Microsoft similarly tried (before being ordered otherwise by the courts) to so thoroughly integrate Internet Explorer into Windows that you basically couldn’t use one without the other. Google seems to be trying to do the same with all their products and Google+. The‘Photos’ link used to take me to my photos on Picasa. This was helpful. Now it takes me to the recent photos that people in my circles have uploaded to Google+. This is not generally useful, and makes it harder to get to my pictures. On pretty much all their products the Share button (which used to take me directly to the option to email or link to an item) now opens a window asking me to post whatever it is on Google+, making it harder to share whatever it is with the vast majority of people I communicate with through other means. I understand that Google wants me to communicate with all of them only through Google+, but as that is never going to happen, designing a product that assumes it has already happened is stupid. There are quite a few of these small things that don’t make a big difference in and of themselves, but take Google’s products, which have traditionally been beautifully designed and implemented one step closer to things I use because I have to. And once you are into “use because I have to” territory, you are Microsoftian.

Computer bugs

We now have enough ladybugs in our window plants that the aphids and white-flies won't last long. However keeping the predators concentrated in the little ecosystem for that long requires some doing. We find these little red happy pills wandering all over the apartment, and occasionally find them congregating in some warm dark niche. They like to be in groups, as their defensive chemicals and colors work better that way. This morning I found them assembling a conclave on and in the wireless router. Warm. Dark. Dry. Connected. Perhaps it is time to supply some hidey-holes in the planters under the plants.


P.S. Yes, I know that they are actually beetles, not bugs, but who has ever heard of computer beetles?

Kudzu Power

For several years I've mused that if only we could find some efficient way to make ethanol from Kudzu (the incredibly fast growing invasive weed that has taken over much of the southeastern US), we could stop importing petroleum, vastly improve many ecosystems and tremendously improve the US economy, all in one swell foop. I didn't take this idea very seriously. It appears that somebody else has. Consider the following abstract from a recent volume of the journal Biomass and Bioenergy:

We determined the amount of standing biomass of kudzu (Pueraria montana var lobata) in naturally infested fields in Maryland and Alabama, USA. At each site, we evaluated the carbohydrate content of roots, stems, and leaves. For a third site from Georgia, we evaluated the carbohydrate content of kudzu roots of varying diameters. Belowground biomass in Alabama exceeded 13 t ha⁻¹, and contained an average of 37% fermentable carbohydrates (sucrose, glucose, and starch). Roots from Georgia of all size classes contained over 60% fermentable carbohydrates. Biomass and carbohydrate levels in roots from Maryland were low compared to plants growing in Alabama and Georgia, producing 5 t ha⁻¹ of roots with 20% non-structural carbohydrate. Stems from Alabama and Maryland contained 1–3% carbohydrates. Based on the yield and carbohydrate content, we estimate wild kudzu stands in Alabama and Georgia could produce 5–10 t ha⁻¹ of carbohydrate, which would rival carbohydrate production from maize and sugar cane fields. If economical harvesting and processing techniques could be developed, the kudzu infesting North America has the potential to supplement existing bioethanol feedstocks, which could be of significance to the rural economy of the southeastern USA.

Further consider this economical processing technique, from next month's PNAS:

Abundant plant biomass has the potential to become a sustainable source of fuels and chemicals. Realizing this potential requires the economical conversion of recalcitrant lignocellulose into useful intermediates, such as sugars. We report a high-yielding chemical process for the hydrolysis of biomass into monosaccharides. Adding water gradually to a chloride ionic liquid-containing catalytic acid leads to a nearly 90% yield of glucose from cellulose and 70–80% yield of sugars from untreated corn stover. Ion-exclusion chromatography allows recovery of the ionic liquid and delivers sugar feedstocks that support the vigorous growth of ethanologenic microbes. This simple chemical process, which requires neither an edible plant nor a cellulase, could enable crude biomass to be the sole source of carbon for a scalable biorefinery.

This will, of course, take longer and be more complicated than anyone currently expects, but I'd be surprised if we don't eventually have industrial scale application of this technology, using many sources, including kudzu.

H1: The universe is 96% aliens.

Most good scientists are deeply impressed by, and even excited about, how little we yet know, how much there still is to find out. Our science and technology are impressive accomplishments, but still very much in their infancy as compared to what is possible.

Imagine, if you will, a tremendously technologically advanced civilization, whose technology just continues to advance. What is the end point? What do they and their technology look like after billions of years of rapid technological advance? I think it is safe to say that we don't know. It clearly wouldn't be anything that we currently have the science or technology to understand. As Arthur C. Clarke wrote, "Any sufficiently advanced technology is indistinguishable from magic." We wouldn't understand this alien technology or what it did any more than Cro-Magnons would understand a wireless modem. This hyper-advanced civilization would presumably have explored ever nook, cranny and loophole in the physical laws that govern the universe and transformed their technology and themselves beyond recognition. We likely wouldn't even recognize it or them as technology or life. We might not even recognize them or their technology as normal matter or energy, although after billions of years they would likely have transformed large parts of the universe itself.

Which brings me back to how little we currently know. Modern cosmology assigns only 4% of the mass/energy in the universe to "normal" matter, the kind we know anything about. 22% is thought to be "dark matter" and the remainder, 74%, is "dark energy." We know little about dark matter and next to nothing about dark energy, except that they are necessary to make our equations work out. Putting to one side the plausibility that the equations are wrong, we are free to speculate on what is going on with that other 96% that we can't see but suspect is there. My preferred hypothesis is that the universe is 96% aliens. Which is a somewhat silly way of saying that other life-forms may have transformed some large portion of the mass and energy in the universe into forms we don't understand. Why should a civilization with the most advanced technology we can pretend to conceive of be expected to leave things as they found them? If conservation of mass/energy is one of the rules they can't get around, couldn't a big crowded universe of aliens be expected to convert a large portion of its limiting resource into whatever form allows them to use this resource most efficiently? Dark matter releases no detectable radiation, and as such leaks no mass/energy. This would also help to explain why SETI can't detect any aliens. A civilization that hoards mass and energy isn't going to go blaring electromagnetism all over the place.

Why under this hypothesis would 4% of the matter be left as boring old stars, dust, black holes and light? Perhaps these things serve some function for the dark aliens. Or maybe we, and everything we can see, are a designated nature preserve with strictly enforced limits on poaching. To try to answer this last question is to move from pure speculation to pure fancy, so your guess is as good as mine.

Scientific publishing after paper

Yesterday I submitted a paper to the Proceedings of the National Academy of Sciences. Like most high profile journals they have strict page limits, and my paper was carefully planned to be just under the limit. I got an email from them today saying that they had recently changed the format in which references are given in the journal, and I would have to use the new, longer, format, and resubmit. I had 65 references, so the longer format made my paper half a page too long, which means I have spent the morning making the paper shorter again.

All this has got me thinking about the future of publishing, as fewer and fewer people read their journals on actual paper. The traditional justification for page limits, and a lot of the other rules journals have, was that it cost the journal money to print the extra pages, or the color, or figures above a certain size. Several journals, including PNAS, have even offered authors the option of going over the page limit or adding colors, but paying the extra printing costs. What happens to these limitation as fewer and fewer subscribers bother with the paper edition, and almost everyone goes straight to the digital version?

We are not near abandoning the paper journal yet. Even journals that start out as internet only, as PLoS did a few years back, often end up printing a paper edition because libraries and some readers (especially the older and more established) really like have a paper journal to thumb through. But this is changing. I, and most researchers my age, read almost everything from a digital copy. Some people print out the individual articles they want to read, others read from a screen. It is rare to see anyone my age or younger reading from a bound print journal, unless of course it is one of those rare journals that isn't available online. It is becoming increasingly common for libraries, to save shelf space and money, to provide access to only the digital copies of the less frequently read journals. There are journals which can only be found online, and I suspect their number will increase rapidly over the next decades. These journals (e.g. JoVE) can be far more multimedia than the old ink on rectangle.

So far, this seems to be having relatively little effect on the page limit. Journal editors want authors to get to the point, even if they don't have to worry about the cost of paper and ink. Journals are increasingly offering the option of accompanying the paper with supplemental online material. Those experimental protocols, that video sequence, the large table and the mathematical appendix can all be posted online, with the permalink listed in the main paper and hyperlinked directly from the digital versions. PNAS allows up to ten such files, and the length of all these add-ons often surpass the article itself, but without busting the page limit.

Annoying as it is to have to go through and find three more lines of text to cut to get down to that page limit, I suppose I should be happy that the page limit will be around long after the page itself has gone the way of the mimeograph. I don't want to have to read as much as all those people would write.

I do think the journals themselves will last, even if only online. Someone has to organize the peer review, and everyone needs to have a limit on the number of places they need to look to find relevant papers. I doubt I will see the day when one can be a successful scientist just by uploading my own work onto my own website and inviting people to read it.

Field Guide to the Birds of whereever I am right now.

Planning a recent trip to Baja California, I decided to buy a field guide to the birds of Baja, only to discover that there is no such book. Ornithologist colleagues suggested I just bring a guide to the birds of Mexico, but that seemed much less than ideal. Mexico has more than a thousand species of birds. Somewhere around 300 of those 1000 have ever been seen anywhere on the peninsula of Baja, and maybe a 150 of those have a moose's chance in Texas of showing up where we where when we were there. I looked through my very old edition of Peterson's Field Guide to the Birds of Mexico (so old they don't even have pictures of all the species because of the cost of printing illistrations), and decided that almost every bird we were at all likely to see was in the much more usable National Geographic Field Guide to the Birds of North America (which in this case means the US and Canada), so I just brought that.
Of the 66 bird species we saw, only one wasn't in the Nat Geo guide, or at least only one I successfully identified, the Grey Thrasher.

But this system of having to have a different guide for each place one goes is just so cumbersome, especially when one goes to a place for which a guide is not available. It's time for field guide 2.0. What I want is an electronic guide that detects where I am and what season it is, then displays a list of species that could possibly be there. This could even be a fairly simple application for an iPhone or PDA. I click on the name of the species I want to see, or the group I want to explore, and I get that page. If I know I am looking at a booby, but don't know which one, I click on the genus Sula. Based on the fact that I am in Baja in winter, I get pictures of a Blue-Footed Booby and a Brown Booby of the brewsteri subspecies.

If I am an ambitious birder and hope to find birds that aren't normally found where I am, I tell the program to be less picky in the list it gives me. If I am a novice who will only notice the species that there are at least a thousand of all aroud me, I can get a more selective list and have an easier time IDing the Yellow-footed Gulls.

This would require no new technology, only someone from one of the several companies who make bird guides to take the data they already have and slap them in a program. But no, instead they want to sell us stacks of bound paper. How 20th century.

(EDIT: Just after posting this it occurred to me that someone might already bo doing this. The closest I can find are "Handheld Birds" from National Geographic and iBird Explorer. They don't yet have any more birds than what appear in field guides to the US and Canada, and they don't seem to have the capability to let your wireless device filter by your location and season, but I hope that will come soon. It is clear at least that bird guides are going digital.)


Here, by the way, is the unorganized list of bird species Iris and I saw on our trip:

La Paz, La Ventana & Puerto San Carlos, Baja California Sur, Mexico Dec. 15-25th, 2008

Birds
1. White-winged Dove
2. California Quail
3. Magnificent Frigate Bird
4. Brown Pelican
5. Turkey Vulture
6. Crested Caracara
7. Merlin
8. American Kestrel
9. Western Gull
10. Gila Woodpecker
11. Cassin’s Kingbird
12. Cactus Wren
13. Northern Mockinbird
14. Gilded Flicker
15. Phainopepla
16. Western Scrub Jay
17. Spotted Sandpiper
18. Royal Tern
19. Ring-Billed Gull
20. Orange-Crowned Warbler
21. California Gnatcatcher
22. Common Ground Dove
23. Common Raven
24 Costa’s Hummingbird
25. House Finch
26. House Sparrow
27. Pyrrhuloxia
28. Hooded Oriole
29. Great Egret
30. Sanderling
31. Bonaparte’s Gull
32. Lesser Scaup
33. Double-Crested Coromorant
34. Green Heron
35. Little Blue Heron
36. Great Blue Heron
37. Snowy Egret
38. Great Egret
39. Cattle Egret
40. Tricolored Heron
41. White Ibis
42. Osprey
43. Red-Tailed Hawk
44. Heermann’s Gull
45. Western Sandpiper
46. Rock Pigeon
47. Mourning Dove
48. Anna’s Hummingbird
49. Blue-Footed Booby
50. Caspian Tern
51. Forster’s Tern
52. Eared Grebe
53. Yellow-Footed Gull
54. Ash-Throated Flycatcher
55. Grey Vireo
56. Verdin
57. Rock Wren
58. Semi-Palmated Plover
59. Long-Billed Curlew
60. Black-Throated Sparrow
61. Belted Kingfisher
62. Ladderback Woodpecker
63. Willet
64. American Oyster Catcher
65. Grey Thrasher
66. Brown Boobie

Better Solar Cells

It is amazing how when one does research into a technology, that technology can get better.

Science 11 July 2008:
Vol. 321. no. 5886, pp. 226 - 228

High-Efficiency Organic Solar Concentrators for Photovoltaics
Michael J. Currie, Jonathan K. Mapel, Timothy D. Heidel, Shalom Goffri, Marc A. Baldo

The cost of photovoltaic power can be reduced with organic solar concentrators. These are planar waveguides with a thin-film organic coating on the face and inorganic solar cells attached to the edges. Light is absorbed by the coating and reemitted into waveguide modes for collection by the solar cells. We report single- and tandem-waveguide organic solar concentrators with quantum efficiencies exceeding 50% and projected power conversion efficiencies as high as 6.8%. The exploitation of near-field energy transfer, solid-state solvation, and phosphorescence enables 10-fold increases in the power obtained from photovoltaic cells, without the need for solar tracking.

Windbelt

This is a really cool invention my friend Brent brought to my attention. It would be incredibly useful in places like PNG where solar doesn't work because it is always raining.