Wednesday, June 28, 2023

SearchResearch Challenge (6/28/23): How can you find a free audio book?

Audible books are life-changing. 


P/C Tomas Anunziata on Pexels.com 

When I go on long road trips, listening to audible books (previously known as "books on tape") are an essential item. They keep me awake and listening, especially when I'm driving those long, straight stretches of highways in the American west that seem to go on endlessly.  

Sometimes I don't want a podcast but just a good old-fashioned novel.  When I need such a thing I often turn to places like Audible.com for recent books and pay a small amount of money for the right to download the audible file. 

But every so often I want to listen to a book that's a classic--Treasure IslandCatch-22, or The Return of Sherlock Holmes.  Is there a simple way to do this?  That's the Challenge for this week... 

1. Is there a simple way to find free audible versions of classic books?  (Such as Treasure IslandCatch-22, or The Return of Sherlock Holmes.)  What's your method of finding these? 

2. Do you know of a public repository of less common books that would also allow you to download the audio files?  (Books like The Autobiography of Mother Jones.

3. My cousin in Stockholm recommended I listen to a book that she really liked:  Män som hatar kvinnor.  But I can't find an audio version of that title in English.  Can you find one?  

This week isn't that hard, but it brings up a couple of good points about searching for something that's not just another web page.  See what you can do with this! 

Keep Searching! 



Wednesday, June 21, 2023

Answer: How to find the best AI-powered search engine of the moment?

How can you find the best tool... 

P/C Dall-E. Prompt "clash of the titans - three giant science fiction robots all fighting together with thunder lightning dramatic skies


 

... to do that search thing we all know and love?  Is there a kind of meta-search tool for search-engines?   

As you well know, there are LOTS of options for doing SearchResearch operations.  So this week had a different kind of Challenge--a kind of meta-search Challenge--a search for the best research tools out there. 

1. How can I find the best AI-powered search tools out there? Is there a way to do this? 

2. What's your favorite AI-powered search engine?  Say why it's your favorite?  

Naturally, I did multiple approaches to this.  Obvious search #1 on both Google and Bing: 

     [ best search engines ] 

This gave multiple lists of search engines, most of which won't surprise you.  Bing, Google, Yandex, Baidu,  Naver (Korean). 

There are a lot of others as well that are less well-known, but rely on conventional search engines to create their index and/or provide their results:   these are search engines like DuckDuckGo (Bing), SwissCows (Bing),  StartPage (Google),  Gibiru (Google).  Their advantage is that they offer privacy protection for your searches.  

And there are speciality topic search engines: Wolfram Alpha (science, math, technology), TinEye (image search), SlideShare (for presentation slide decks) etc.  

As we've noted before, there's a nice Wikipedia page with a List of Search Engines. As with most Wikipedia pages, it's fairly up-to-date, so that's a good resource to check every so often.  Put that on your list of things-to-check once a month or so.  

Are these "AI-powered"?  If we would have asked that question 12 months ago, the answer would have been NO, but of course, this is the Year of AI, so now the answer is YES, of course. 

If we're more specific, say "AI chatbot interfaces to search engines," then the list gets smaller.  Google and Bing have such frontends, but so do You.com and a few others.  

Thing is, many of the general purpose chatbots also act as a general purpose question-answering tool.  In a sense, they're ALL chat front-ends. You can ask any of them to (prompt) "give me a list of the top ten search engines with chatbot interfaces."  That's effectively asking any chatbot to act like a classic search engine.   

As you know, questions like "how can I find the best X" (for any kind of X) involves testing, evaluation, or some kind of basis for recommendation.  The key question is how do you evaluate a chatbot search engine?  

If we ignore measuring things like quality of the conversation, but just stick to how accurately the chatty search engine answers the questions, it's a different ball game.  

Microsoft and Google are reviewing and rating their new gizmos and looking at what the raters are saying, but it's an incredibly tough problem--they're trying to evaluate their ability to answer literally any question.  (It doesn't help that Google's own internal raters are complaining that they're not being given enough time to actually do the research needed to give an accurate rating.  See: Business Insider article, April 4, 2023.) 

So I searched for: 

     [ compare accuracy of chatgpt and bard answers ]  

I'm only asking about ChatGPT and Bard based systems.  (There are other LLMs out there, but I don't expect to find comparisons just yet.  Maybe in a year or so.)  

Since everything is changing so quickly, I then limited the search to just results within the past month.  You can do this with a tool (click on Tools then Last month) or you can modify the query like this: 

     [ compare accuracy of chatgpt and bard answers after:2023-05-20 ]

This query leads to a bunch of fascinating results.  You can look at each of the articles in detail--here's the list of the top five articles I found with short summaries:  

Lifewire:  Bard is the clear winner, but ChatGPT is better on productivity. 

Zapier: Bard is better generally, but ChatGPT is a better writer.  

ExpressVPN: Tie

Duplichecker: Very close, slight edge to ChatGPT for ability to use more writing styles. 

TechRepublic: Bard is free and the training data is up-to-date (not true of ChatGPT).  

Bottom Line:  With this quick survey, as of this moment, it looks like Bard is edging out ChatGPT... but this might change at any moment.  And, for coding and productivity applications, ChatGPT is the winner there.  But again... change is constant.  

So.. what's my favorite chatty search engine?  Answer:  Neither--or both.  I never do anything without running the prompt on each. I like some aspects of one (the writing style or the ability to reframe a long prompt) versus the other depending on the task.  I have to agree with the authors of these articles.  When I've tried coding tasks, ChatGPT usually does a better job. (And yes, I almost always run my tasks / queries on another chatty search engine like You.comPerplexity or Scite.ai 

But given how quickly everything is changing, I have no doubt that these results will change with time.  Stay tuned.

It seems like every week there's yet another LLM chatbot put out into the public domain. So we have a job to do: keep tracking what's going on in the field and let us (the community) know what you're finding.  Please feel free to add a comment to this post if you find something new and fantastic--it's an easy way to share the knowledge.  

When you're looking for the latest update on LLMs and search, remember to use the before: and after: operators.  That way you can search month by month for the latest updates.  (You'll find, for instance, that Neeva, which is cited in many of the articles as a new, hot, exciting search engine, actually shut down on June 2, 2023.)  


SearchResearch Lesson

It's pretty obvious this week: 

1. You have to keep your eyes and ears open for the latest and greatest.  There's not going to be a clear winner anytime soon.  Check back here with SRS--I'll do my best to keep you in the loop about which chatty search engine is (currently) the best. 

Keep searching!  (And keep sharing your knowledge...) 





Friday, June 16, 2023

Another cautionary note: Why you need to fact-check everything an LLM-based system tells you

 It was annoying, but necessary.. 



.. I needed a new laptop, so I hied myself down to the local Apple shop and picked up a fancy new laptop.  Not a problem--I've used Macs for years (except when I worked at IBM, when the IBM PC was the axe of choice), so I know the drill.  I know how to shift all my stuff from old-Mac to new-Mac.

But I'd forgotten how many settings needed to be copied over as well. There's the wifi, the default font size, the settings for folder appearances, etc etc.  

Then I noticed a new behavior that was driving me crazy.  Whenever I was working in Chrome, the system would add an extra period after a closing parenthesis.  It looked like this: 

 (blah blah blah.). 


Why would it add an extra period?  Don't know.  After fooling around for a bit, I figured out that the problem was not Chrome adding a period after the parenthesis, but after I typed two spaces.. then it would add an extra period  (I learned to type back in an era when two-spaces were the default.)  

So I did the obvious thing and asked Bard:  

  [ how do I get Chrome to stop adding an extra period after two spaces? ] 

Here's what it told me.  See that line with the big red arrow in the image below?  Yeah... that's bogus.  That line and everything after it is just purely made-up.  There IS no "Advanced" option under Chrome settings, so everything after line 3 is just hallucinated and utterly bogus.    


If you follow these instructions, you'll see this after you select "Settings": 


Note that there's no "Advanced" option here.  

Of course, this could be advice about a previous version of Chrome (but wouldn't that be relevant information to include here?).  

So it was with no little surprise that I learned today that Google is telling its employees to be cautious in their use of Bard.  They're more concerned about people leaking sensitive information, but the general policy of Fact Check Everything still applies here.  

After a little more conventional poking around, I found that the actual way to fix this problem is by changing a setting in the Mac OS!  (See this article for how to do it.)  That is, the problem had nothing to do with Chrome at all!  

Instead, the problem really is that the systems we use are composite systems.  An app like Chrome sits on top of the Mac OS, which in turn uses more systems below it.  The behavior you, the user, sees is the compilation of everything below. 

In this case, the "adding a period after two spaces" thing is part of the text handling system of MacOS.  Chrome can layer its own behaviors on top of that (such as redefining how Control-F works, but that's another story).  

What this means for you as a SearchResearcher is: 

1. Behaviors you see in your computer might be caused by any of a number of settings.  Don't be quick to blame the system you see on top (Chrome, in this case).  The troubles might be caused deeper in the stack. 

2. Once again, Fact Check everything.  Even things that look like simple documentation might be pure fabrications.  Check before wasting your time.  


Keep searching.  


Wednesday, June 14, 2023

SearchResearch Challenge (6/14/23): How to find the best AI-powered search engine of the moment?

We have to do something differently...

P/C Dall-E. Prompt "clash of the titans - three giant science fiction robots all fighting together with thunder lightning dramatic skies

 

The world of search is changing dramatically every day.  Bing launches a new thing, then Google Search launches a new thing. New LLM engines seem to be popping up every day.  What is a SearchResearcher to do in a world full of change, claims, counterclaims, and an endlessly shifting searchable ground?  

The good news is that there are LOTS of options for doing SearchResearch.  The bad news is.. there are a LOT of options, and I'm not sure which is best, for what purpose, and how to be effective at the search.

So this week we've got a different kind of Challenge--a kind of meta-search Challenge--a search for the best research tools out there. 

1. How can I find the best AI-powered search tools out there? Is there a way to do this? 

2. What's your favorite AI-powered search engine?  Say why it's your favorite?  

I'm hoping that we'll get enough comments to hear the pros and cons of different tools.  Here's a short list to get you started: 

* Google.com 

* Bing.com 

* Scite.ai 

* Perplexity.ai

There are more, and even additional tools that stretch the definition of "search engine," but I'd love to hear from you about what you found and HOW you found it! 

I'll post a couple of short comments this week as I work through things.  I'd love to see what you use in your day-to-day searching, and what's not working for you.   

Keep searching!  

 

Thursday, June 8, 2023

Answer: Did they really burn Roman statues?

 Did they? 

A scene from 18th century Rome by Giovanni Battista Piranesi. Veduta della Piazza di Monte Cavallo (View of the Piazza del Quirinale with the Statues of Horse Tamers in side view), from Prianesi's Vedute di Roma. Note the marble columns just ready to be rolled away and repurposed.  These statues are still standing in Rome, notably not-broken up, although there's now an obelisk between them.  

The Challenge this week was to determine if this comment found in a recent article in the Atlantic Monthly magazine (My Night in the Sistine Chapel, by Cullen Murphy) is actually true or not... 

"For centuries, the bountiful supply of ancient statuary unearthed in Rome had been burned for lime to make mortar..."  

The image of people just tossing works of art into a kiln to make quicklime just killed me.  This is the kind of thing that makes you say "really?"  (And I'll have more to say about this response later this week.)  

1. Is that sentence true?  Once upon a time did people in Rome just burn ancient marble statuary in order to make quicklime for construction purposes?  

In short, the answer is YES, lots of people did this.  Here's how I found out.  

The obvious search to answer this question is: 

     [ burn statues ancient Rome to make lime ] 

This led me to a Medium post (by Guillame Depres, an art historian and author of a book about lost / destroyed artwork over time) who wrote that:  

"...The city was taken in 410 when the Goths of Alaric sacked and looted Rome...  A second sack by the Vandals of Genseric in 455 lasted 14 days, making the name Vandal a byword for destruction.

But can looting by Goths, Vandals or any other invader account for the disappearance of the vast amount of stone that comprised the [many] temples and the marble statues too numerous to count? The date of the inventory listing the treasures of Rome, the 4th century AD, is of great significance. It appears that mid 4th century the Roman administration in charge of quarrying ceased activity, so no new building was built at the end of the Roman Empire with freshly quarried stone." 

Later in that same article--

 "Not only had the Romans lost vast amounts of the Ancients’ knowledge, they also lost their ambition. It was easier to dismantle the accomplishments of the architects and artists of antiquity instead. So the ancient glories of the Eternal City became a gigantic and convenient quarry.

The age of the builder was replaced by the age of the lime burner, as marble, cooked in an oven, became mortar for the marble cutter who dismantled blocks of temples..." 

"...“the statues lie broken in fragments, ready for the lime-kiln, or are made use of as building material. I have seen many used as mounting-steps, or as curbstones, or as mangers in stables” or in a “foundation wall which is built entirely of fragments of excellent statuary”. Further “in the walls and foundations of an old house, eighteen or twenty portrait-busts of emperors were discovered … fragments of an exquisite statue of Venus built into a wall … a very great number of fragments of the most beautiful statues, which had served as building materials”.

High quality marble was particularly sought after as “..many torsos and statues discovered in digging cellars used to be thrown into the kilns, especially those sculptured in Greek marble, on account of the wonderful lime which they produced."  

Depres quotes widely and refers to  the 1899 book by Rodolfo Lanciani, The Destruction of Ancient Rome: A Sketch of the History of the Monuments, and Pagan and Christian Rome, and The ruins and excavations of ancient Rome. (You can see it in full-view for people who want original sources. I spent a happy hour reading through this 19th century book, which is a delight.)  

The author, Lanciani, I discovered, was a highly regarded Italian archaeologist who worked primarily on the history of Rome with lots of publications and work to his credit.  So Depres has a great source for the content of his article.  

BUT... 

While this is intriguing, it's only one source.  How could we find another source that would confirm or contradict this story?  

Of course, I could go further down into the SERP, but as I read through the various hits, I learned that a section of Ancient Rome had become so famous for burning marble into lime that it was called the Calcarium, which means a "place for burning to make lime."  

That's such a technical term that it's probably going to be used only by real ancient Rome specialists, as such, it's a useful query term that will likely point out pages written by real experts.  So... 

     [ burn marble calcarium ] 

led me to another fascinating book, The Ancient Monuments of Rome and Their Use as Suppliers of Remnants for the Construction of New St. Peter’s Basilica: Building Activity in Rome during the Renaissance.  

Using Google Books, I was able to find a PDF of this title, but I had to do a little digging to figure out where it came from.  Why?  Well... 

It turned out that a Google Scholar search didn't tell me much--the citation had no useful metadata, it looks like this: 


Yes, it's a citation, but how do I evaluate it as a source if all I know is the author and date?  

If you look at the PDF, at the bottom of the page you'll see this as a footnote on the first page: 


That's the place where authors often put the metadata for a chapter that's part of a collection forming a book.  This tells us that the original is in German (entited "Perspektiven der Spolienforschung") and comes from something called the Berlin Studies of the Ancient World.  

That sounds good, but what is it? A quick search to show the connection between this title and the Berlin Studies would be:  

     [ Berlin Studies of the Ancient World "the ancient monuments of Rome"] 

This gives us a result to the Berlin Studies page, where we see this text with our target in the middle: 


(See update below)  As you can see, it's part of a multi-volume set, all of which are learned academic articles about the archaeology of the ancient world.  So the correct citation should be something like: Bernhard Fritsch, Die Dekonstruktion antiker Räume und die Spolienverwertung beim Neubau von St. Peter in Rom, Berlin: Edition Topoi, vol. 63 (2018).  That is, the "Edition Topoi" is a collection of articles that form a kind of book, where each chapter comes out independently.  This is the kind of thing that drives librarians (and Google Scholar) crazy--it's not quite a book, but not quite a journal, it's somewhere inbetween.  

Update (June 12, 2023):  Regular Reader Miguel points out that I messed this attribution up quite badly. In fact, the link above ("Die Dekonstruktion antiker Räume..") points to a real, honest-go-God book (and not a 'collection of articles').  

Here's my mea culpa and an explanation:  There are two relevant documents, one an article by Fritsch from 2017, and a book from 2018 on the same topic.  Here are the two (correct) citations:  

Article: Fritsch, B. (2017) The ancient monuments of Rome and their use as suppliers of remnants for the construction of new St. Peter’s Basilica: building activity in Rome during the Renaissance. In​ ​S. Altekamp, C. Marcks-Jacobs, P. Seiler (a cura di), Perspektiven der Spolienforschung 2.​ ​Zentren und Konjunkturen der Spoliierung, 335–56. Berlin: Edition Topoi.

BookDie Dekonstruktion antiker Räume und die Spolienverwertung beim Neubau von St. Peter in Rom. (Trans: "The deconstruction of ancient rooms and the use of spolia in the construction of St. Peter's in Rome") Berlin: Edition Topoi (2018)  

Obviously, the 2018 book recycles a lot of the material from the 2017 article.  That's fine, people do that all the time--the book extends this work by quite a bit.  But in my haste to get this post out, I made a mistake and thought that the article was part of a multi-volume thing.  My (glaring) mistake. 

Topoi was a Cluster of Excellence funded by the Excellence Initiative of the German federal and state governments from 2007 to 2019. They published a lot of work in this area--see their list of books on the topics of archaeology, Rome in particular, and the interpretation of artifacts.   

In any case, this article by Fritsch agrees with the Depres article: After the fall of Rome, it was easier and faster to burn marble statues, columns, and building facades than to mine the limestone from quarries far away.  The market price for construction-grade quicklime was higher than that of statuary, so into the kilns they went, beauty be damned.  

Of course, I wanted to see the cred of this author.  By checking on the author, it's easy to learn that Bernhard Fritsch is a respected author of many papers on Roman and Italian antiquities, publishing in high quality journals.  (You could search for his papers on Google Scholar, or look for a summary of his work.)  

That's what I did to learn the answer.  

But my friend Jamie Teevan at Microsoft wrote to me with an answer she got from ChatGPT-4.  The answer she got was pretty good.  The ChatGPT-4 answer was: 

Hello, this is Bing. I’m glad you found something so astounding to share with me.😊

To answer your first question, yes, it is true that ancient statuary was burned for lime in Rome and other places in the Mediterranean world. This practice was common in late antiquity and the Middle Ages, when marble was more valued as a source of mortar than as a work of art.12  Marble can be burned above 840°C to convert the calcium carbonate into calcium oxide, commonly called quicklime, which is an essential ingredient to make concrete.3 The burning of marble was such a vast industry in Rome that the district where the kilns were concentrated came to be called the calcarario, meaning "mortar district."1

As for your second question, I’m not sure what other ancient material you have in mind, but one possible example is papyrus. Papyrus was used by ancient Egyptians, Greeks and Romans for writing and painting, but it was also used for making baskets, mats, sandals, boats and even fuel. Papyrus was made from the stem of a water plant that grew along the Nile River. It was cut into thin strips, soaked in water, pressed together and dried to form sheets. Papyrus was widely traded and exported throughout the ancient world, but it became scarce and expensive after the Arab conquest of Egypt in the seventh century CE.

Some other possible examples of ancient materials used for surprising purposes are:


  • Mummies: Ancient Egyptian mummies were sometimes ground up and used as medicine or paint pigment in medieval and early modern Europe.
  • Spices: Ancient spices such as cinnamon, pepper and cloves were not only used for flavoring food and drinks, but also for preserving meat, masking bad smells, making perfumes and incense, and treating various ailments.
  • Glass: Ancient glass was not only used for making vessels, jewelry and windows, but also for creating optical devices such as lenses and prisms.

I hope this answers your questions and satisfies your curiosity. If you want to learn more about any of these topics, you can use me to search the web. Just type your query in the chat box and I’ll do my best to help you.😊


Overall, this is pretty good.  However.  The first citation (superscript 1) points to a Reddit article by Alkibiades415, who seems to be an informed historian, but I don't know anything about their background, so I don't know how authoritatively the comment should be taken.

The second citation (2) points to the same article we found above (by Depres), so that seems pretty good, and we already know it's good quality.  

The third citation (3) on the other hand points to a syndicated copy of this SearchResearch blog post, which is a bit... odd.  It's not a wonder that it matches in a search on this topic, but you shouldn't cite the original post as support for your argument!  

So, as we've discussed earlier, LLMs are an interesting source of insight, but you still have to verify everything...   


2. (Just for fun..)  I know of at least one other surprising use of ancient materials for the most prosaic of purpose--can you figure out what that other ancient material is (was)?  

On the other hand by the same authority... ChatGPT-4 did a great job figuring out answers to this fairly open-ended question.  

I posed this Challenge question because I had in mind a very specific "ancient material" used for prosaic purposes. I was thinking of mummies because, strangely enough, they've been used for all kinds of strange (and prosaic) purposes.  

Here's what I did.  

     [ uses for mummies ] 

which leads to an Encyclopedia Britannica article on 7 Surprising Uses for Mummies.  I probably don't have to tell you that the E. Britannica is fairly reputable, so when they say that Egyptian cat mummies were sold as fertilizer, it's such an odd thing that I fact-checked the encyclopedia.  Sure enough, there was a thriving business in excavating cat mummies from vast animal cemeteries in Egypt for just their fertilizer value.  This was so common that the practice was satirized in the British magazine Punch (1890).  


Another remarkable use of mummies was for locomotive fuel and as the base color in "mummy brown," a popular paint color among the Pre-Raphaelites.  Go figure.  

Linked image of mummy brown paint.  Link to Harvard Art Museums.

The value of something is determined by its value in the market of the time.  Context, as they say, is everything.  If you're scrambling to make a living while surrounded by thousands of Roman sculptures, the value of that statue is higher as lime than as art.  Sic transit gloria mundi.  

But for SRS work, I was impressed that ChatGPT4 also suggested glass as an option (I hadn't thought of that), but not impressed by the suggestion of spices.  This highlights a characteristic of LLMs as question-answering systems--they might not pick up on the implication that we were looking for "ancient materials (then re-used) for prosaic purposes."  

I think in future SRS episodes we'll have to spend more time talking about how to use LLMs (ChatGPT, Bard, and similar) in productive ways.  

SearchResearch Lessons

1. Check your sources. Where did the article appear?  Who wrote it?  What's the reputation of the publisher and what's the reputation of the author?  In these cases, the publishers and the authors checked out.  But it's important to ALWAYS check. 

2. Using a specialty technical term can find specialty articles.  In this case, I learned the word "Calcarium," which was handy in finding articles specifically about burning marbles to make lime.  This leads to another lesson... 

3. As you read, note the speciality terms--both for understanding AND for search purposes.  You can often pick up terms and language that will lead you to very precise queries.  I actually write down such terms in a notebook for just such a reason.  (And if I haven't mentioned it recently, be sure to take notes as you search!)  

4. LLMs can be useful for searching complex or amorphous topics.  I was impressed by ChatGPT4's answer, even if it was sketchy in the details.  But if you think of LLMs as a friend who has vast knowledge, but that you always have to fact-check, then that's about right.  We'll talk more about this in posts to come. 


Keep searching!