I am already hyper aware of the way Facebook can really exacerbate confirmation bias and polarize politics and sort of give us a weirdly tweaked perception of our lives in a kind of idealized mirror view of ourselves, so an article warning about that filter bubble created by such parsing of news feed items according to FB algorithm was less alarming to me than the one that popped up as related to that one. THAT one goes a bit more into how some of these personalized results can get rather insidious in ways we don’t even realize until it is much too late on even more supposedly neutral or unbiased sites, because the biases that are built in to the filter aren’t even recognized until something gets personally noticed as a really fucked up result. And they won’t even know what bias created it to fix it! .
(read about that some more here: http://www.washingtonpost.com/news/the-intersect/wp/2015/03/23/what-you-dont-know-about-internet-algorithms-is-hurting-you-and-you-probably-dont-know-very-much/)
I think that anything that shapes our thinking by culling the information we get exposed to needs to disclose the process by which that information is selected to be excluded – as well as being very up front and open about how much is actually getting cut being very prominently displayed up front when the information is requested,
But I do want the burden of filtering out the garbage to be left to me to do myself by default and allow others to ask me if I want to be optimized by analysis of the past behavior rather than it being applied as a filter before I get to see what I might want to see and they don’t let me see it because I didn’t know I wanted to see it before I saw it – you know?
Me, personally – my best optimization algorithm for personalized results
One of the most important factors in the optimisation of results is the question of for whom exactly the results are supposedly optimized to benefit. If it is supposed to be for ME, well then how does the algorithm determine what I want NOW based on what I wanted before?
This is obviously interpreted incorrectly by Google’s search function, since they did away with the “search within results” option that used to appear when the damn thing was far more useful and relying primarily on boolean logic and the searchers particular skill with keywords and their own ability to reason and apply that boolean logic to hone the results as they saw fit. I was very good at that for my purposes, and for finding what others wanted to find better than they could find it themselves.
My success was greater because of a particularly useful set of skills to that purpose that is purely contextual knowledge of words and people and how they might habitually behave even if it isn’t how they logically should. Such as a broad vocabulary which allowed me to search common words that most would search as well as less popular but highly specialized and specific words in nuance or synonymous meaning that differentiated the results from less desirable but accurate matches returned by the more common references, coupled with some insight to people’s personality quirks which allowed me to search for things based on the mistakes people would commonly make if they were not native to the culture or outside the usual demographic (think grandparents trying to google a video game to get for their grandkids and not knowing the specs of the jargon which would most likely get them what they would really want, and avoiding some of what they might actually get that they really wouldn’t want to give to children at all that might have adult content without their realizing it – OOOOPS!)
This is not just because I can recognize patterns in behavior, but also because I can intuit or reason out WHY those patterns exist and adjust accordingly when the pattern alone doesn’t generate the correct result. That is a particular analytical ability that the algorithms cannot match without being told to do so by someone who has realized that the results were poor in the first place. FEEDBACK to the algorithm about its results and their usefulness is not immediate enough to adjust for in a subsequent search, but a person knows immediately when they didn’t get what they wanted. Without the ability to know how the information is filtered, even someone who would be able to alter their search to find a better answer cannot do that because they have no idea what variables are in the equation and how they are weighted to produce the bad results they got and it ends up an exercise in futility and experimental frustration.
Another plus in my column as internet sleuth and hunter of all things obscure extraordinaire is my own technical understanding – and the awareness of how the people who programmed the tech might think differently than people who USE it can also be a factor – and my somewhat unusual ability to interface with both people and machines with pretty equal fluency, when many are usually much stronger with one or the other the way someone is right of left handed. This means that I not only understand how the boolean logic terms and ordering of the search will affect results depending on how things are grouped and that it matters where you put the OR’s AND’s and NOT’s much more than people realize when you are talking to a computer program as opposed to your English professor. The computer will not adjust for what it understands you REALLY meant when you input something that you thought you wanted the way your prof might, it will just give you what you actually asked it for, which can often differ GREATLY from what you wanted to get.
And that is really the thing I miss most about that Search within results action, in that it was much easier to start with what you KNOW was actually working and then be able to eliminate the things that were not in a sequential manner of paring down by subsets. It was the process of combining exact boolean reasoning with the more nuanced variables to adjust the formula to get custom results for each and every different reason for each search and was a very efficient method of separating the wheat from the chaff regardless of however that wheat variety changed from search to search. An algorithm sees wheat as WHEAT, and sorts accordingly, without the ability to alter the grain’s texture or choose different strains according to whether its to be made into something else as an ingredient or whether it is to be of the best quality for its own standalone value for direct consumption, nor does it account for the fact that the premium wheat may be much more expensive and impractical for someone to use if they intend only to pound it into a flour to make some kind of natural adhesive paste that only needs to be non-toxic and not necessarily tasty.
Reasons for what we want what we want change as well as what we want, but the formulas are not equipped for guessing our motivations when we ask it a question at all. And people are often not even always aware of their own reasons for wanting something enough to tell an algorithm that might be clever enough to ask for that information to aid its selection even if there were one that dared to do that. I could do better than many people could do for themselves with that old unadulterated by personalisation boolean slave that Google search used to be because I am intuitive and curious and highly empathetic and well educated and articulate and have pretty good reasoning skills for critical analysis as well as broad exposure to various cultures and diverse cross sections of the population outside of my own comfort zone and familiarity, Awareness and objectivity are not the same thing, and though I may not always be able to tap into both at once, I do have the understanding that sometimes I might confuse them and what that might do to my perception.
An algorithm just is never ever going to get that fucking personal. People can’t even get that personal. But maybe one person can get that close to themselves to find what they really need if they look for it. I don’t mind overwhelming amounts of garbage I don’t need coming at me when i first sit down to look at it, because I was really good at knowing how to tweak my inquiry just so and eliminate the majority of the unwanted stuff myself right away, So in an overwhelmingly impossibly sized haystack with a mountain of needles buried in it, I’d essentially be an electromagnet that only attracted needles made of a particular alloy – leaving me to search for a needle in a moderate sized pile of needles if I wanted one that was a rare or unusual blend of elements, But the algorithms can only make a really unwieldy pile of needles that is just as impossible to go through one needle at a time or maybe an unwieldy amount of smaller specific haystacks with needles that I have to do specific searches in each one to find what I sek making that equally inefficient.
Finding a really obscure needle is easier my way. So I want to be able to search for information MY WAY. Cause I;m better at it than any algorithm that can ever be devised by anyone anywhere ever…I am a big girl and would prefer that I be allowed to decide for myself what kind of results I want to see without the helping hand of personalization. It seems like those results eventually have the actually opposite effect they intend by trying to filter according to whatever it assumes about me or why i want to find anything.
I LIKE to grow and change, so it only makes sense that I will not be content by the answers selected and filtered by any function that might generate any algorithm to do my choosing for me. since it lacks prescience to see the future. That actually is a pretty good way to ensure that I don’t change by repeating the past.
Because by definition, things that don’t grow or change don’t LIVE.