Slashdot Mirror

Bennett Haselton writes "Evgeny Morozov's forthcoming book To Save Everything, Click Here describes how an overly helpful 'kitchen of the future' might stifle the learning process and threaten culinary innovation. True, but we could certainly do better than the current state of how-to directions (in cooking and most other subjects) that you can find today on Google. I suggest that the answer lies not in intelligent kitchen technology, but in designing an algorithm that would produce the best possible how-to directions -- where the 'best' directions are judged according to the results that are achieved by genuine beginners who attempt to follow the directions without help." Read below for the rest of Bennett's review.
Editor's Note: This article was not intended as a full review, but rather a commentary on one point in the book. The author's actual review of the book will appear in March. To Save Everything, Click Here author Evgeny Morozov pages 432 publisher PublicAffairs rating 9/10 reviewer Bennett Haselton ISBN 1610391381 summary Argues that we badly need a new, post-Internet way to debate the moral consequences of digital technologies

Evgeny Morozov's new book To Save Everything, Click Here (due out in March), about "the folly of technological solutionism", is that rare animal: a book I would recommend to everyone even if I disagree with about 2/3 of the conclusions in the text. The arguments in the book didn't always change my mind, but they made me reformulate many of my own arguments in the other direction.

In most sections of the book, Morozov attacks the beliefs of "solutionists" who believe that a particular program or algorithm can solve a social program. Usually, I thought his criticisms of a given algorithmic "solution" were spot-on. But I often found myself thinking of a different algorithm that I thought would solve the problem much more effectively than the one Morozov was critiquing. This, naturally, could be construed as missing the point of the book. However, I'm prepared to defend any of the alternative algorithms that I came up with, or bet money on how it would fare in the real world. I'll have a full review of the book when it's released, but I think many of Morozov's argument are interesting enough to deserve an article in their own right.

For example, Morozov describes a new kitchen technology that guides would-be chefs through the process of preparing a meal, by illuminating pathways on the kitchen floor to show the cook where they're supposed to walk next, and then using laser pointers and visual aids to guide them through what they're supposed to do when they get there. If you want to know how to expertly carve a fish, for example, the ceiling-mounted lasers will trace out the exact cuts that you're supposed to make on the fish's skin. The description sounds like a parody of what people think the Big Bang Theory geeks would like their kitchen to do for them.

Morozov argues, not unreasonably, that "[t]o subject [cooking] fully to the debilitating logic of efficiency is to deprive humans of the ability to achieve mastery in this activity, to make human flourishing impossible and to impoverish our lives," and that "deviating from recipes is what creates culinary innovations." Well that's one of the 1/3 of his arguments that I agree with. Besides, if you can afford the cost of a laser-guided kitchen just to cook meals for yourself, you could probably use the same amount of money to take a professional cooking class, order takeout every day to tide you over until you know how to make decent stuff on your own, and still have money left over. If you're using it instead to try and cook to impress party guests, how's that going to work? If you're making the food where your guests can see you being guided around by lasers, they're going to think (correctly) that you don't know how to cook, and if you're making the food in a back room where you're out of sight of the guests, you might as well order takeout and have it smuggled in through the back door.

On the other hand, Morozov says in his next paragraph: "In a world where only a select few could master the tricks of the trade, such 'augmented' kitchens would probably be welcome, if only for their promise to democratize access to this art. But this is not the world we inhabit: detailed recipes and instructional videos on how to cook the most exquisite dish have never been easier to find on Google."

That's where he lost me. I have vastly different views on this, which can be summed up in three points:

• The qualify of most "how-to" instructions aimed at beginners, judged by the results they produce in the hands of actual beginners, is far worse than most people believe.

• Moreover, for reasons I'll describe later, the incentives created by the free market in general (and Google in particular) more or less guarantee this result: How-to directions exist that cover nearly every human activity, but most of the directions are not particularly good.

• I have an idea for a different algorithm (surprise!) that Google, or any other similarly positioned web titan, could use to change the incentives of web publishers, leading them to write how-to instructions that would produce much better results when followed by actual beginners.

The morass of cooking how-tos on the web are a good example. Partly from always having other things that I'd rather learn, and partly from being perfectly happy eating lots of plain fruits and vegetables (good for your health, but not for your cooking skills), I had survived to early adulthood hardly knowing anything about real cooking. Being a decently smart person, I figured that made me well suited to judge the effectiveness of the countless cookbooks written "for people who don't know how to cook". Because I firmly believe that if you follow a set of directions precisely (or, if the directions are written ambiguously, then if you follow some plausible interpretation of each step in the directions), and the result doesn't come out as predicted, then it's the directions that failed, not you. If another set of instructions would have produced better results, then those directions are better. This is not rocket science, but many cooking directions in cookbooks and on the Internet are glaringly missing key pieces of information that would have made the directions better, by the above definition.

Now, I understand the importance of experimenting and deviating from recipes and tailoring things to your own tastes, but I think that has to come after you've produced an edible dish that you can use as a baseline. I make scrambled eggs a little bit differently every time -- curry powder, mussels, capers, tabasco sauce, blue cheese (just not all in the same pan, please) -- but the only reason that's possible is because the simple directions for plain scrambled eggs actually work. When I say that most cooking directions don't work, I mean that if you follow them precisely (but without any prior cooking knowledge), they don't even get you to the baseline of an edible result that you can then use as a jumping-off point to try your own variations.

The odd thing about cooking is that of all the people whose cooking I liked so much that I asked them where they learned how to cook, all of them said that they learned from an in-person instructor (usually a family member); I have yet to meet any really good cooks who learned their skills from written recipes or web videos. This suggests that the learning materials on the Internet are falling short. (By contrast, I know plenty of people who have learned PHP programming or similar skills out of a book.)

And from my experiences helping out friends in the kitchen who had more cooking experience but who were trying to follow a particular recipe, it seemed that their most valuable skill was knowing the crucial parts of the recipe that were missing, or wrong. And then they would use their non-beginner knowledge fill in the missing steps or make the necessary corrections as we went along. With the current mediocre state of most cooking directions out there, that's surely a useful skill. However, it does mean that you could make most recipes produce much better results in the hands of a beginner, if you simply fixed all those parts that were missing, or wrong.

Take, for example, my misadventures making jalapeno poppers. Going to a friend's Super Bowl party, I figured that jalapeno poppers would be an easy thing to make, with just under 200 how-to videos on Youtube and about 600 matching recipe pages on Google, most of them calling for only four ingredients. How hard could it be?

Well, there are two important things that should be in every jalapeno popper recipe, or the recipe is doing more harm than good just by being out there on the web. One is that when you're slicing and handling the raw jalapenos, you have to wear gloves, or the capsaicin in the jalapeno -- which is also the active ingredient in pepper spray -- will leave a burning feeling on your fingers that lasts for about the next 24 hours. (If you touch your eye with your finger, you might even have to go to the emergency room.)

The other indispensable piece of information is that to make the jalapeno poppers edible, you have to remove the seeds and the white ribs from the inside -- not just the white center of the jalapeno (which slides out easily), but the white part of the ribs, which have to be scraped off of the outer wall (a grapefruit spoon works great, otherwise a paring knife or a regular sharp knife will do). Most recipes do tell you to remove the seeds. But the white ribs left inside the jalapeno are just as hot, and if you don't cut them out, the finished product will have a hotness that's too overpowering to taste anything else. (This video shows how to do it right.)

So what's the problem? Here's a table listing the first 10 Google matches for "jalapeno popper recipes", rated according to whether they contain those two must-have pieces of information that a beginner would need. (If the directions said to "devein" the jalapeno or "remove the membranes", I gave it an "Almost" in the second column -- because a first-timer is likely to think that this refers to removing the white center of the jalapeno, and not realize that you also have to remove the ribs attached to the edges. I'm being strict here, because it would have taken almost no effort for the recipe writers to be clear about this, and if you don't do that step correctly, you will have to throw out the finished product.)

Recipe source Tell reader to wear gloves? Tell reader to remove jalapeno ribs? Food Network (Emeril Lagasse) No Almost (instructions say "membranes removed") AllRecipes.com No No Food.com No No KraftRecipes.com No Almost (says to remove "veins") InspiredTaste.net Yes. (Sort of. The directions end halfway down the page, and then another set of written directions starts from the beginning. That's confusing, but I'll give it to them.) Yes. (In both sets of directions. Good job guys!) ThePioneerWoman.com No Yes Epicurious.com No Almost ("devein") About.com No (not counting the comments section, where someone warns other readers to use gloves because they burned their hands following the directions) No RecipeGirl.com No. (This is weird: gives tips on how to neutralize the stinging capsaicin once it gets on your hands, but never actually says to put gloves on.) Almost ("seeds and ribs") JalapenoMadness.com No No

Videos scored a little better, if you're generous and give full credit to any video that shows the scooping out of the jalapenos to include the ribs attached to the sides, even if the verbal directions don't spell that out precisely. Here are the ratings for the first 10 Youtube matches for "jalapeno poppers recipe":

Source (Youtube user) Tell viewer to wear gloves? Tell viewer to remove jalapeno ribs? allrecipes Yes No bettyskitchen No Yes PrincessDiana161 Yes Yes MudRFunR Yes Yes cookingwithcaitlin1 Yes Yes Michael Hultquist (Jalapeno Madness) No No BarbecueWeb No No kooktocook No No Adley Stump No No thatsletitia No No

In most of the videos that didn't explicitly include the step about putting gloves on, the cooks themselves were not wearing gloves. What did their hands feel like later?

eHow.com does have a helpful page about how to treat capsaicin burns from handling jalapenos. Perhaps that's their penance for the fact that half of their 'jalapeno poppers' recipes don't tell you to put gloves on.

If you could have made poppers based on these incomplete instructions, because you knew to put gloves on or to scrape the ribs out, good for you -- you possess the background knowledge to fill in the parts of the directions that were missing, or wrong. But that doesn't do the real newbies any good.

I went to this trouble because I want to beat you over the head with the crucial fact here: Most directions suck. They suck not just in absolute terms (burning your hands, or the mouths of people who eat the jalapenos with the ribs still in them) but they especially suck relative to how easily they could have been fixed. There is no excuse for putting up a recipe for jalapeno poppers that doesn't tell the reader to put gloves on, or that only tells the reader to "remove the seeds". And I've run into the same phenomenon over and over -- whether looking for directions on how to lower memory consumption of a web server, or how to get stains out of a carpet, or how to replace a 12V direct-current power supply with a cartridge of 8 AA batteries in series -- where not only did the directions not work, but I later found out that they could have worked if the author had simply added one or two key pieces of information.

However it seems that almost everyone believes that the quality of directions on the web is much higher than it actually is -- where, by "quality", I'm talking about the results that would be achieved by a beginner following the directions. (If I had asked you, "Where can I find a good recipe for jalapeno poppers?", is there about a 100% chance you would have said, "Google"?) I assume people overestimate the usefulness of all the how-tos out there, for two reasons: (a) they glance at the directions but don't try them themselves, so they just assume the directions work; or (b) they already know how to do the task being described, so when they read the directions, their brain automatically fills in the missing steps or makes the necessary corrections. That doesn't mean the directions would work in the hands of a true beginner.

Unfortunately, the quality of the directions on the web, is perfectly explained by the incentives created by Google. If there's any niche in the how-to space that is not already filled by some article on the web, an author can easily grab some extra web traffic by writing the first page about that topic. For a popular topic like how to make jalapeno poppers, there's enough traffic going around that dozens or hundreds of authors can put up their own how-to pages and each collect just enough web traffic to make it worthwhile. Thus, every "directional" niche will be filled, and some will be filled to overflowing.

Within a particular niche, however, there's not much incentives to make the directions particularly good -- where "good" means "produces good results when followed by someone with no prior knowledge in this area". Whether your directions work or not, they'll attract about the same level of traffic from Google. Even if the author later realizes that the insertion of a few key steps would make their instructions better, there's no incentive for them to do it -- that's not going to make your how-to page rise up in the Google rankings above the other pages on the same topic.

Which brings me to my proposed solution. It would take a company with a giant pre-existing web presence to pull it off (not quite on the level of Google, but at least an eHow or a Food.com). But it would take almost no maintenance on the part of the company themselves, once the process was put in place.

To incentivize people to create instructions that actually work, a given how-to guide would go through three phases:

1. After the directions are written, genuine newbies (recruited from the web site's usual visitors -- people who just want to learn something new in an area where they have no prior expertise) attempt to follow the directions and tell the author about any problems they ran into, or steps in the directions that seemed ambiguous. If the author thinks some reader is just being an overly nit-picky moron, they're free to ignore their questions and suggestions, but they would do so at the risk of their directions faring poorly in the next phase.

2. Once the initial wave of corrections and clarifications is finished, the directions are put into a pool marked "Ready to be rated!", where they are rated by the next group of genuine newbies who attempt to follow them. Each reader rates the directions simply: If they followed the directions and got the result they expected, then thumbs up, otherwise, thumbs down. If multiple readers spot a mistake or an omission that somehow got missed in the first phase, then the author can make the necessary changes and start the second phase over. (To prevent the author of the directions from "gaming the system" at this stage, the volunteer newbies should be selected at random from a large pool of people who sign up saying "I'm game for learning how to do anything new." If you let people self-select to go to the directions and rate them, then this enables the author to stack the deck by having all of their friends go to the page and give their directions a high rating.)

3. Finally, once the directions have reached some acceptably high percentage of positive ratings, they get released into the general pool of directions/how-tos/recipes of which the site can promise, "80% of newbies were able to follow these directions successfully." If the system works -- and if the volunteer readers in step #2 are representative of the skill level of the site's general readership -- it should be expected that most readers should be able to follow the directions and get good results at that point.

Almost all of the "how-to" directions that I've read, on any topic, could have benefited from being put through the wringer as described by the steps above. It's not merely that I think this algorithm would produce good directions; it's that my definition of good directions is precisely those directions that would pass the test in step #2.

As for what incentivizes the authors to produce directions that make it through this process, perhaps the hosting site could split the ad revenue with them from the pages containing the author's directions. Perhaps the hosting site could just reward them with a link from the article to the author's professional home page. Or maybe people would happily submit the instructions for free if it went towards a non-profit repository of helpful information, a la Wikipedia. (The huge difference from Wikipedia though, is that if you're an expert on George Washington, it's easy to write a good article about George Washington; but if you're an expert on cooking, that makes it hard to write a set of cooking directions that would fill in all the blanks needed by a beginner. Hence the multi-step vetting process above.)

It's tempting to think this is process would be "overkill" for a simple recipe, but that fails to consider the magnitude of the time savings when multiplied across the hundreds or thousands of people who will read the information over the course of its lifetime on the web. If the author spends an extra 10 minutes on the instructions to clarify things in such a way that saves just 1 minute of reading time for the average reader, when that 1 minute of time savings is multiplied by hundreds of readers, it's clearly an overall time-saver. (What disgusts me about the jalapeno popper recipes is that the authors could have saved me a whole day of painful burning on my fingers, if they had just taken 10 seconds to include the step about putting on gloves -- that would have been an overall time-saver even if only one person had read the recipe.)

So Morozov was right that we don't need laser-guided kitchens guiding us through the algorithm of carving a fish, but we should consider that an entirely different kind of algorithm could change everything for a beginning cook, or a person trying to learn any other skill from scratch. The Star Trek kitchen in To Save Everything, Click Here makes for an easy target for Morozov's argument, but that kitchen technology is hardly making enough inroads to threaten cooking as we know it -- I'll bet you'd never heard of it until this article. Bad directions, on the other hand, are so ubiquitous that we've accepted them as a part of our way of life, and we've all but forgotten to think how they could be made better. Like Robert Kennedy, I see people looking at their capsaicin-burned hands and their inedible jalapeno poppers with the ribs still attached and asking, "Why?", and I imagine eHow.com lining up newbies to critique their recipes until each recipe achieves a high rating from beginners based on the actual results that they got, and ask, "Why not?"

You can purchase To Save Everything, Click Here from amazon.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.