Portraits of Wildflowers

Perspectives on Nature Photography

Time again for mountain pinks

with 51 comments

Flowers and bullet-like buds of Zeltnera beyrichii on June 18th in Cedar Park.
Thinks to Kathy Werner for tipping me off to the location.
(In return I tipped her off to the location of some bluebells near there.)


◊◊
◊◊◊

A few days ago I finished reading Douglas Murray’s excellent book The Madness of Crowds, in which he pointed out something I’d begun noticing, too: the hits that come up in response to certain searches on Google are biased. Murray gave several examples, one of which was that when he searched for “straight couples,” many of the pictures that came up in Google Images showed gay couples. His book is from 2019, so I tried that experiment for myself last week to see what sort of results I’d get in mid-2021.

The top row of hits I got for “straight couples” contained seven pictures. The first showed a lesbian couple. The second showed a gay male couple. The third showed a male-female couple. The fourth showed a lesbian couple. The fifth showed a male-female couple. The sixth and seventh both showed lesbian couples. In summary, only two of the seven pictures in the top row matched the search string “straight couples.”

It’s practically impossible for a set of hits so different from the search string to come up by chance. To understand why, imagine all the pictures of couples out there on the internet; billions of them have been posted. Now imagine that you searched for pictures of couples without specifying any particular kind of couple. Using the estimate that 5% of couples are same-sex, I did the calculations to find out how often a random grab of seven pictures of couples would yield an assortment with five gay couples and two straight couples. The arithmetic shows you can expect that to happen only 0.14% of the time, or approximately 1 out of every 700 times. And remember, that’s without specifying what kind of couple you’re after. The fact that I searched specifically for straight couples makes the 5-gay-and-2-straight result I got much less probable than the already tiny 0.14% we’d expect if we didn’t specify the kind of couple.

The only conclusion possible, in fact the one Douglas Murray came to, is that Google is cooking the books—and since Google is a search engine and not accounting software, cooking the books means rigging the search algorithm to distort reality. And this from the company whose original motto was “Don’t be evil.”

Oh, and just in case anyone feels an overwhelming ad hominem urge to label Douglas Murray homophobic for pointing out what he did about Google, he happens to be gay.

© 2021 Steven Schwartzman

Written by Steve Schwartzman

June 28, 2021 at 4:27 AM

51 Responses

Subscribe to comments with RSS.

  1. A beautiful photo and an interesting story. Google has been selecting what information I find first for months if not years. My grandmother once talked of her PRIDE in her queen and country. Now google the word PRIDE and see what you come up with. I think she would be rather sad if she were still alive!

    Cathy

    June 28, 2021 at 6:36 AM

  2. I love these pinks, they almost appear to be in motion –

    beth

    June 28, 2021 at 7:11 AM

    • You’re not alone: I’ve long gotten that feeling of centrifugal motion from mountain pinks.

      Steve Schwartzman

      June 28, 2021 at 8:17 AM

  3. Very nice pink flowers. Everyone’s business goal now is to show up in a Google search. I thought of doing the experiment myself but every search results in a flood of ads everywhere including e-mails.

    Alessandra Chaves

    June 28, 2021 at 7:42 AM

    • No question but that Google and other companies track our internet movement on our computers and phones. Big Brother turned out to be big tech.

      Steve Schwartzman

      June 28, 2021 at 8:19 AM

      • It tracks everything I do! It’s convenient service so we put up with it!

        Alessandra Chaves

        June 28, 2021 at 8:20 AM

        • We pay a big price for that convenience, not in dollars but in surveillance.

          Steve Schwartzman

          June 28, 2021 at 8:22 AM

          • I know, and who’s to guarantee the security of all this data? Even if the law could always protect us agains legal misuse of this information, how about hackers?

            Alessandra Chaves

            June 28, 2021 at 2:55 PM

            • Right, and we’ve heard about the many large-scale hacks that have taken place recently, even at institutions you’d assume would have very robust security. Who knows how many other hacks have taken place that we haven’t heard about, and that the hacked institutions might not even realize happened?

              Steve Schwartzman

              June 28, 2021 at 3:58 PM

              • I feel like I have been forced into giving personal information at more and more places. Last attempt was my clinics wanting me to open a health record online so I can have access to my lab tests, talk to my doctor etc. I don’t want my health information online. I like the file folders in the doctor’s office just fine. There is no more doctor-patient confidentiality. It’s all online.

                Alessandra Chaves

                June 28, 2021 at 6:38 PM

                • Even if you didn’t participate, I imagine your doctors’ offices would still increasingly digitize their records and store them in “the cloud.”

                  Steve Schwartzman

                  June 28, 2021 at 6:58 PM

                • I don’t participate but I think that you’re right it’s in some cloud and yet I never gave them permission. I was also not asked.

                  Alessandra Chaves

                  June 28, 2021 at 8:21 PM

                • Welcome to our world in 2021.

                  Steve Schwartzman

                  June 28, 2021 at 8:37 PM

  4. I submit that the search terms do not understand ‘straight’ in that search term

    Elisa

    June 28, 2021 at 8:49 AM

    • Ok wow apple you changed my words to gibberish again, after i hit enter! The search engine, the ‘mind’ doesn’t understand straight and couple together. Straight is a specific kind of a line 🙂

      Elisa

      June 28, 2021 at 8:51 AM

      • I think “straight” in the context of couples is pretty well understood. But even if you eliminate the word “straight” from the search string and just search for “couples,” you still would have only a very tiny chance of getting the assortment I got.

        Tangentially (speaking of lines), you may wonder why we redundantly say “straight line” when all lines are by definition straight. The answer is that English originally used “line” in the more general sense of ‘curve,’ so “straight line” designated a certain kind of curve. The original ‘curve’ sense of “line” is still with us when we speak of the sleek lines of a speedboat or of a streamlined locomotive.

        Steve Schwartzman

        June 28, 2021 at 11:17 AM

  5. Steve, I understand the point that you are trying to make, but I question your approach. The Google search engine is text-based–unless someone specifically tags a photo, as I do with my wildlife subjects, Google has no way of know the subject of the photo. I would suggest that very few straight couples tag photos of themselves with the label “straight couple,” so more often the term “straight couple” is used in only very limited circumstances. Why did those particular photos come up? They came up because they were part of the titles of articles comparing gay couples and straight couples, such as “What straight couples can learn from gay couples,” as Google makes abundantly clear by posting links to those very articles. Your point about bias in the search algorithms may be well-taken, Steve, but I find your evidence in this case to be really weak and your approach flawed–because it is based on the faulty assumption that the content of a photo causes the Google result rather than the associated text.

    Mike Powell

    June 28, 2021 at 10:12 AM

    • Hi, Mike. You make a good point about Google’s search algorithm being text-based. What I don’t know is whether its search engine can also “read” pictures for their contents. In any case, I’d noticed, as you did, that some of my hits came from pages with titles like “What straight couples can learn from gay couples,” which could account for some or a lot of the skew.

      On the other hand, I just tried a different search engine, Bing, to see what sort of results I’d get. If Google’s search algorithm is text-based, then presumably Bing’s is as well. A search for “straight couples” turned up nine photographs in the first row of Bing Images. Six of the pictures showed straight couples, versus three same-sex couples (all three of which came from pages with text like the “What straight couples can learn from gay couples” that we’d both noticed.) In summary, Bing largely gave me what I searched for, and Google largely gave me the opposite. How would you account for such different results?

      Steve Schwartzman

      June 28, 2021 at 11:46 AM

      • I don’t know how the different algorithms work, but one major difference is possibly the different sample size (Google returned 468 million results for the search “straight couples,” and Bing returned 3.18 million results). Why is there such a difference in the size of the “pool” of results? I don’t know why, but they may be using different sets of data. The biggest unknown, though, is the weighting and prioritization of results, i.e. which ones rise to the top and why. There is an entire industry devoted to SEO (search engine optimization) that tries to figure out how to manipulate the results so that their clients are near the top of the results. In the case of the Google results, for example, what is it about the 2019 article “What straight…” that caused it to rise to the top of the Google text results? I looked through the first ten pages of results from Bing and didn’t find the article at all. Why? Again it is hard to know. My blog post today noted the difference between the Google and Bing results for “Red-footed Cannibalflies,” a topic that has no obvious political bias. On Bing, one of my postings is the third result listed on the very first page, while on Google, I show up on the third page of the results because of the tag that I used for my WordPress postings (and the link takes me to all of the postings I did with that tag, not to the posting by that name.) My takeaway is that different search engines produce different results, but I think that it is premature to jump to any conclusions about the reasons on the basis of a very limited sample size.

        Mike Powell

        June 28, 2021 at 12:58 PM

        • I’ve read about people trying to game the system to get their product or service or advocacy position to show up among the first hits on Google. Once in a while a post of mine has showed up among the first few hits on Google, but I forget what I searched for. It may have been a certain native species in central Texas that not a lot of people out there on the Internet would have been writing about.

          Regarding the difference in the number of hits that Google and Bing returned, you’re right that we’d need to know how the data is obtained. If both companies’ hits come from representative samples in the “population” of what’s out there on the Internet, then 468 million versus 3 million shouldn’t matter. I seem to recall from elementary statistics that once you get past a random representative sample in the thousands, a further increase in the sample size make practically no difference in the accuracy of the results. The key, of course, is to have a random representative random sample, and I have no way of knowing if that’s true for Google or Bing.

          I’ll agree that more research is needed to bolster or refute Douglas Murray’s inference about bias at Google. People who did that research would need a lot more technical knowledge than I have about search engines and the Internet. That said, there’s at least some prima facie evidence that something other than chance is at work at Google. (And there are those occasional troubling instances I’ve read about of Google employees talking about and even advocating the manipulation of results in the service of a favored cause.)

          Steve Schwartzman

          June 28, 2021 at 1:30 PM

          • Totally agree, Steve, that Google, for business and other possible reasons, is far from neutral in serving up results and, of course, is collecting information about us. I therefore try to be a little cautious when reviewing results of any searches.

            Mike Powell

            June 28, 2021 at 1:45 PM

            • If only the “news” media and politicians exercised your recommended caution, rather than so often jumping to conclusions about incidents before investigators have examined the evidence. As a recent example, take the June 19 vehicle crash at a Pride event in Florida that killed a man. Almost immediately Fort Lauderdale Mayor Dean Trantalis described the crash as a “terrorist attack on the LGBT community.” It was soon revealed, however, that the driver of the vehicle was himself a participant in the Stonewall Pride Parade and a member of an associated men’s chorus. According to NBC: “[O]n Sunday, Trantalis clarified that while [his original] assertion had appeared ‘obvious’ to him and others at the time, facts uncovered by investigators showed the crash appeared to be an accident involving a truck that ‘careened out of control.’”

              Steve Schwartzman

              June 28, 2021 at 2:11 PM

  6. Great floral carpet as always, Steve! What you came across in your research on the Google search engine was a real eye-opener.

    Peter Klopp

    June 28, 2021 at 11:52 AM

    • I’m not sure this one qualifies as a floral carpet, Peter, given that you’re seeing a closeup of a single plant. On the other hand, if we look strictly in visual terms, this picture fills the frame with colorful and shapely element the same way a photograph of a broad wildflower colony does.

      As for the big tech companies, their behavior isn’t always ethical.

      Steve Schwartzman

      June 28, 2021 at 12:51 PM

  7. This has the appearance of a slight zoom while exposing. Probably just an optical illusion but your mention the other day of trying zooms made me think that it might.

    Steve Gingold

    June 28, 2021 at 2:28 PM

    • I took this picture at 1/400 of a second, so I’d have to have been one super-speedy zoomer. The flowers and buds in a mountain pink plant tend to form a dome, which in this case I shot straight down at. That’s what gives the illusion of centrifugal force at work, when in fact everything was stationary.

      Steve Schwartzman

      June 28, 2021 at 3:54 PM

  8. Duh! I tried to find information about ‘Canna indica’, but found mostly information about ‘Cannabis indica’. That is what ‘normal’ people look for, right?

    tonytomeo

    June 28, 2021 at 6:44 PM

    • I just did a Google search for “Canna indica” and, at least on the first four pages of hits, got not a single one for “Cannabis indica.”

      Steve Schwartzman

      June 28, 2021 at 7:46 PM

      • That’s pretty excellent! I have found it difficult to research this and other species of canna. Could this be a regional observation?

        tonytomeo

        June 28, 2021 at 8:07 PM

      • Hey! I just tried it and got the same results (after the advertisements for dispensaries). What is up with that?! I’m certainly not complaining.

        tonytomeo

        June 28, 2021 at 8:08 PM

        • Don’t know, but it’s good the search is working properly for you now.

          Steve Schwartzman

          June 28, 2021 at 8:35 PM

          • Well, now I don’t need it! Well, there is another species of Canna that I want to track down.

            tonytomeo

            June 28, 2021 at 8:52 PM

  9. Nice

    Asmin

    June 28, 2021 at 8:08 PM

  10. That strikes me as hilarious that so many gay couples come up when you search for straight couples. I can only agree with you, Google must be cooking the books, but I cannot see a purpose nor a harm in it.

    melissabluefineart

    June 29, 2021 at 8:06 AM

    • One speculation is that the purpose is to make it seem that gay couples are much more common than they actually are. Look at this Atlantic article from 2012:

      https://www.theatlantic.com/politics/archive/2012/05/americans-have-no-idea-how-few-gay-people-there-are/257753/

      Steve Schwartzman

      June 29, 2021 at 8:13 AM

      • I suspect that is the case. The thing is, don’t straight people know they are straight? What is the fear?

        melissabluefineart

        June 29, 2021 at 9:08 AM

        • “Fear” may not be the right word. There are many adherents to traditional belief systems who feel homosexuality is wrong. For example, there are some 1.9 billion Muslims in the world, and from what I’ve read, “standard” Islam does not accept homosexuality. Many Christians don’t, either. Taken together, those two groups account for a large share of humanity. You may disagree with them, but of course they disagree with you.

          Steve Schwartzman

          June 29, 2021 at 9:32 AM

          • Of course. And it has been established that homosexuality is genetically determined. I suspect the percentage of the population carrying that gene has not changed. What has changed is acceptance, permitting these people to come out of the closet. It certainly isn’t the only thing I disagree with religion on. Christians can thump their Bible all day long but if they’d actually READ the thing maybe they’d notice that Jesus embraced all peoples.
            All that aside, I’m confused about why you even bring this stuff up on your blog. Where are you coming from on this?

            melissabluefineart

            June 30, 2021 at 9:33 AM

            • I’m very worried about the increasing amount of censorship and the slanting of the news to favor certain ideologies. So is Douglas Murray, the author of the book I mentioned. If you searched for asparagus and got a lot of hits about broccoli, you’d be justified in assuming that, for whatever reason, the search engine is doing something to promote broccoli. You may say that’s a trivial example, but Google and Facebook have been doing that sort of thing to push ideological and political viewpoints.

              Steve Schwartzman

              June 30, 2021 at 10:02 AM

              • Ok, now I understand.

                melissabluefineart

                June 30, 2021 at 10:33 AM

      • The article seemed to suggest that it isn’t homosexuality itself the author objects to, but same-sex marriage. I don’t understand this. I know that economists get frantic when the birth rate drops, but have they looked around lately? We as a species are grossly over-populated. We desperately need a lower birth rate. I can think of no other reason to object to same-sex marriage, can you?

        melissabluefineart

        June 29, 2021 at 9:14 AM

  11. I do love the mountain pinks, particularly when they decorate roadcuts. One of their great advantages is that they’re willing to show up in tough conditions, although they often escape notice because of it. I’ve learned to look over the edge of roads as well as ‘up’ to find them. It’s interesting how their color can vary, too. I suppose it has to do with soil and other conditions, but I’ve seen them as deeply saturated as these, and the lightest pink.

    There’s no question that Google’s been cooking the algorithms, especially in the past couple of years. I’ve learned to use alternative search engines, including internal search engines at JSTOR and other journal-related sites for certain kinds of information. Otherwise, it can be page six or seven before I get a return related to my topic — if I ever do.

    shoreacres

    June 29, 2021 at 8:22 AM

    • Tough conditions indeed: around here mountain pinks often grow in caliche or even seemingly right out of rock surfaces. This year I haven’t yet seen any on or below their usual cliffs along Capital of Texas Highway. Whether the February freeze suppressed them there for this year or whether they’re just late remains to be seen. As you noted, mountain pink flower color varies a fair amount, as does that of horsemints.

      I think more and more of us have been noticing Google’s biased results. Big Brother, thy name is Google.

      Steve Schwartzman

      June 29, 2021 at 8:31 AM

  12. […] June 18th, after photographing some mountain pinks I’d been tipped off to, I stopped at nearby Cypress Creek Park and found to my pleasure that […]

  13. Big thanks to Kathy for tipping you off … those flowers are gorgeous

    Julie@frogpondfarm

    July 1, 2021 at 2:24 PM

  14. […] across the other species, T. scaposa, and took advantage of the find to make a portrait with some nearby mountain pinks offering their contrasting color in the […]

  15. A beautiful explosion of pink!

    Lavinia Ross

    July 2, 2021 at 9:25 AM


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: