Common sense

Don Y snipped-for-privacy@foo.invalid wrote: |-----------------------------------------------------------------------| |"I am increasingly amused -- to the point of dismay! -- about how | |little common sense is noted in "answers" provided by tools. | | | |Search engines, of course, are notoriously bad at allowing for | |criteria refinement -- the more terms you add (increasing specificity),| |the more results you get!" | |-----------------------------------------------------------------------|

Cf. that quoted comp.lang.ada post timestamped Date: Sun, 18 Sep 2022 12:57:25 +0200

- |-----------------------------------------------------------------------| |"[. . .] | | | |> I am very curious what else exists in the Ada ecosystem -- in | | particular in terms of messaging support such as websockets and | | others. | | | |There exists multiple implementations of websockets in Ada as well as | |other network protocols. | | | |I cannot suggest searching the Web, because search engines deteriorated| |beyond any usability. Just ask what you need here. | | | |-- | |Regards, | |Dmitry A. Kazakov | |

(S.

Another point, in favour of AI, this time. The search results from search engines are intentionally crafted to *not* show any pertinent answers in the list of search results, so that you are forced to actually click on a link to see more. (Often to discover that it still doesn't give you the info you were looking for.)

The AI-generated blurb often *does* provide the desired answer without needing to click through.

I expect this will be temporary, until everyone learns to spoil the game.

Jeroen Belleman

This is deliberate. If you search for 'thing', it first shows you what it thinks are the best matches (for you or those which make the best revenue for them).

But tech companies have decided that people don't like being told 'that's it, I didn't find anything else' (because you might leave the site and go somewhere else, and eyeballs = revenue) so what they do is show you junk that may or may not be tenuously connected to what you searched for. You get a thousand pages of search results but only the first handful are actually related to your search.

The trick is to spot at what point the results fell off a cliff and stop there. When you realise this is what's happening it's quite easy to identify.

That's possibly the opposite problem, that the primitive on-device search in the GPS is not fuzzy enough to handle variations. eg if the store is coded in its DB as "Pennys" then perhaps "Penny's" doesn't match.

Theo

Search engines (in the sense of Google, DDG, etc.) are different beasts; a much wider domain and range.

But, a product (e.g., GPS) can provide "assistance" in specifying destinations by constraining the destinations to those that it recognizes. This accelerates data entry as well as eliminating the "Invalid destination specified" error phase of unconstrained input.

In the US, there is a retailer named "J C Penney". Most folks likely refer to it as "Penney's". Or, misspell it as "Penny's". As *both* are valid destination names, there is no way for the user to realize his spelling mistake *or* incorrect name usage.

And, the device sees this as "valid" input ("Hey, maybe the guy actually WANTS to go to Pennys! Someone must as it is a valid destination in my internal database!")

Likewise, "The Home Depot" is typically referred to as "Home Depot" despite the actual name being "The Home Depot". But, again, there is a valid destination called "Home Depot" (that is NOT the destination intended as "The Home Depot").

Again, no way to alert the user to his mistake as it's NOT a mistake in the eyes of the GPS!

However, in each of these examples, the *intended* destinations are within a few miles of my location. Not 1200 or 1400 miles from here!

When the destination found is displayed, none of the details on the map make any sense. You only realize it is incorrect ("not desired") when you notice the distance field (off on the side) indicates more than 1000 miles and a few days driving time!

"Common Sense" would suggest the results offered in these cases are likely NOT what the user sought -- simply because they are so far away even if you ignore the fact that past driving history suggests purely "local" driving -- and, often included destinations (based on their GPS coordinates) that match similar entries in the database.

If I type in EXACTLY the description of an item FROM A COMPANY'S SITE, I expect that to be the FIRST result. Not one of 73,000!

Similarly, if the URL is something like:

I would NOT expect a large number and variety of items that are completely unrelated to "long-description-of-the-item-sought".

The goal of a vendor is to make a sale. Sure, they want to slip in a few "related" items to entice you to make a different/additional purchase.

But, not 73,000 of them.

And, they should make damn sure the item you likely seek is included at or near the top of their results -- lest you go elsewhere.

Pages are cheap. You can add "these might ALSO be of interest to you" to make it clear that they aren't as good of a match to the search criteria.

That makes sense for generic searches. But, when you are being very specific (using the terms that the site has used to identify your item), you expect better.

It goes beyond that.

To their credit, they ignore punctuation -- and even whitespace! But, if the proper spelling is "Penney" (not plural!) and the proper NAME is "J C Penney" -- but it is colloquially known as "Pennys"/"Penneys" (I wonder how many folks would remember the second E in the name?), there is no easy way to accommodate this in a device with limited resources.

Similarly "The Home Depot" is known as "Home Depot" -- to everyone BUT the GPS!

When these "colloquial" names happen to collide with other valid names in the device's dataset, you get results -- that are "correct" in the eyes of the device yet wholly incorrect in the eyes of the user.

Of course, a *web* search will readily identify the desired site.

However, the device WILL know that the results are not local (in these two examples, the results returned by the GPS were for locations

Again, recall the user is interacting with the device IN A VEHICLE while primarily concerned with DRIVING. And, viewing the results on a tiny screen that is trying to convey lots of information, all of it inappropriate for his intended goal! (No, it's not

I can just see the *engineer* who designed the functionality, feeling self-righteous because he produced the "correct" answer (for the input given), despite it being 100% useless.

We use the "egg timer" in our microwave oven for short interval timing -- baking cookies, pancakes, etc. (this because the timer in the oven is ghastly, hard to interact with, doesn't allow specifying fractional minutes, etc.).

Type "45" and you get 45 seconds. Nope, pancakes aren't cooked fully. Let's try 60. Still no good. How about 90? No, just a bit more. OK, 100 should do it!

Wait! That was just 60 seconds, not 100! (Oh...)

And, 160 is actually 2 minutes. 180 is 2:20, etc.

Attempting to be too clever just adds to confusion.

I don't see that at all.

If I narrow down the search with -bad_keyword then all the bogus hits are usually eliminated from that unwanted source.

Where I do see a problem is +wanted_keyword doesn't always work :( (sites I know to contain the last keyword added don't always show up)

Try adding successive terms to this Google search:

Yorkshire +castle +BEST +Howard -lake

By default search terms are logically or together of the keywords if you don't add the prefix or surround a phrase with quotes.

The first 10 hits will generate advertising revenue for the search engine. That's what it is there to do; you are the product, not the customer.

Go to "your favorite retailer's website". The interface is far less "sophisticated" than a search engine like google.

Some make a good effort at trying to "understand" what is being sought. E.g., if I go to the USPS website and type a very long identifier into their search box, it "knows" that I am likely specifying a tracking number for a parcel and produces those results. If I type "forever stamp" I don't expect to see any tracking information.

If I go to my local library's site and type "Hours", I get hits for a film called "Hours", a book called "Hours", a film called "The Hours", books called "His Hour", "By the Hour", etc. Nothing about "Minutes", "Days", etc. Chances are, the earliest results are most likely the desired results.

If I go to Target.com (a department store, here), I get 432 results, with "The Hours" as the 15th result -- behind: Magic Hour The Blue Hour Every Hour Until Then Twenty Four Hours a Day The 24th Hour The After Hours The Witching Hour Blue Hour (not the same as The Blue Hour, above) The Hour I first Believed Our Hours Are Married to Shadow Hour of the Pumpkin Queen Happier Hour One Hour for My Priest The Secret Hours The Hours !!!!!!!!!!!!!!!!

Is there some reason "Magic Hour" is a better result, in their mind? DESPITE MY SEARCH CRITERIA APPEARING EXACTLY AS SPECIFIED IN THEIR RESULTS?

Other sites just throw results at you and leave it to you to sort through THEIR mess.

If Target wanted to sell me that book, I would likely have given up before scrolling through several rows of pictorial results (4 per row).

Is there any guarantee that it will be in the fourth row? Tenth?

Other results:

The Seven Husbands of Evelyn Hugo The First Time I Saw Him The Good Girl Effect Tourist Season Reckless etc.

(and some 15 more pages) "The Seven Husbands of Evelyn Hugo" *almost* matches... there's "The", an 'H' in "Husbands", an 'o' in "of", a 'u' in "Hugo"... just missing the final 'r'!

(Sorry, I should have been cleared, here. "Hours" produces a list of store hours -- a reasonable assumption. So, I searched for "The Hours" as the library had told me it was a valid book title)

At a *store's* web site, the goal is for the store to sell a PRODUCT that they have in stock (as known by the literal description of the item on THEIR web page).

They gain nothing if they point me at products that are not what I expressly sought. All 73,000 of them!

They don't write their own websites, they pay an 'expert' to fill in a template provided by one of the big players. The 'search' feature is ultimately controlled by the major search engines, not by the store.

Some websites are fixed and the client has to pay the webmaster every time they want to update it - so they don't. Even if they have the option to update the website themselves, to reflect their current stock, they soon discover it is a huge chore for smaller businesses with constant stock changes, so they give up and the information will be wrong anyway.

It's all painting by numbers these days. I have had several hand-written customised websites taken off me and handed to 'experts' who replace them with content-free template rubbish with huge graphics that looked good to the client. At least one of my clients had the decency to come back to me and ask me to reinstate his old website as the 'improved' one was useless.

My microwave oven has a dial marked with two different sets of 'minutes' markings (in different colours and fonts to clearly distinguish them). One is the time in minutes during which microwaves are delivered to the food, the other is the time in miuntes it will take to actually deliver this energy when a lever is set to 'Defrost' and the magnetron is pulsed on and off. The timer motor is connected to the magnetron circuit, so it runs only when energy is actually being delivered.

It is simple to understand, simple to use, reliable and extremely cheap to make. It also has real bell that goes 'ping!'.

Ours has independant settings for power level and time.

There are some specialty settings (like "defrost" which accepts a *weight* of the item being defrosted). Or, cooking settings for different types of food (Sauce, Hot Beverages, Snacks, popcorn, etc.)

We bought a microwave for mother-in-law ~40 years ago. At the time, the big choice was "numeric keypad" vs. "dial". I prefered the former because it was likely more reliable (no moving parts) and much easier to clean. But, the "dial" was far more intuitive to her.

The scale was non-linear -- you could dial fine increments of time for small durations (e.g., 10 second intervals up to a minute or so) and coarser for longer periods (minutes).

The oven is a different beast with far more options:

- one or two independant ovens

- bake, broil, convection, dehydrate, etc.

- delay start

- two types of cleaning cycles (quick vs. thorough)

- "Sabbath" mode, "demand mode" (tie in with public utility

- time/date setting, wifi settings, etc.

I.e., you can see how there would be appeal for a "universal" interface. But, no one thought it through in terms of actual use model.

Or, the different types of interactions that one COULD experience. E.g., if you are in the middle of setting <something> and the oven calls for attention -- perhaps because a timer has expired -- the SINGLE UI gets confused leaving you with "Power Off" as the only real way to get to a known state (and then restart whatever was happening at the time).

I can just see the engineer pitching how clever the "universal interface" could address all of the needs of the oven/user. But, no one thinking about how PRACTICAL those interactions would be. Do you really want to have to go through all these steps just to turn the oven on??

We were discussing how AI is scaring all the CE/EE grads, the coders, which makes sense to me. Coding in a precedural language is like spinning and weaving and sewing clothes by hand.

But our issue is whether AI can replace circuit designers. I think not. There are too many possible circuits for a small number of words to invent one. And the side issues of flexibility, manufacturability, thermals, packaging, all that.

As you note, as we think about a design in progress we should be willing to change our minds about everything, including whether we even want to do it.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

So, 1980s or so.

Hmm. What was the given rationale for the no-tools law? Or regulation?

Never had to replace a magnetron, but I took the prior microwave apart and removed the magnetron assembly for some project that never happened.

Joe

Someone has the "responsibility" for the functionality of the site. Someone signs off on whether or not it meets their requirements (if they stated them improperly, then they're at fault).

I have rescued "Google search appliances" (typically 1U servers specialized to handle "local" searches).

The companies in question aren't small or "of limited budgets". From Wikipedia:

"As of 2024, Target is ranked No. 32 on the 2022 Fortune 500 list of the largest American corporations by total revenue. As of 2025, it operates more than 2,000 stores throughout the United States."

with 440,000 employees and net revenue of $106B.

Clearly, other sites work well so the flaw is in the design (specification) of these other sites.

Likewise, the design of our microwave oven, oven/stove, GPS, etc. People put in decision making positions that clearly don't have the right mindsets to make good *use* decisions.

Theo <theom+ snipped-for-privacy@chiark.greenend.org.uk> wrote: |---------------------------------------------------------------------| |"[. . .] | |[. . .] You | |get a thousand pages of search results but only the first handful are| |actually related to your search." | |---------------------------------------------------------------------|

(Google repeatedly changes how it works so the following might not still apply. I performed this experiment before the Year 2019.) I used Google for a search. Google boasted that it found thousands of results, but after I went through hundreds thereof (i.e. a small proportion of these boasted thousands), Google showed me a note that it refuses to list most of these search hits. This note was accompanied by the absence of the Next button arrow icon at the bottom right. (S.

Microwave ovens differ based on:

- internal volume

- maximum power level

- duty cycle algorithm

- "macro" features (presets for common items)

- accessories (e.g., temperature probe, popping corn detector, etc.)

- build quality

Ours has a "beverage" button that is great for reheating hot beverages. It assumes a ~6 oz "dose" and allows you to indicate 1 or 2 "doses" are required (my "tea cup" is 16 oz so I hit "BEVERAGE 2" if I poured it and forgot to drink it or "BEVERAGE" if I've consumed about half and it needs to be "refreshed"

Spaghetti sauce has another setting that neatly fits with the size containers in which I store my sauce. Defrosting from frozen takes some time as it uses a lower power level to do so.

There is a chart of common foods and snacks printed inside the door to let you specify the type of food to cook or reheat.

I *think* it listens for the popcorn to stop popping. (we don't use microwave popcorn)

And, can monitor the internal temperature of a turkey that is being defrosted, etc.

And, a button to enable/disable the "keyclick".

Of course, the general purpose timer and clock (smart enough to realize no need for AM/PM!)

By far, the most used controls are the "add 30 seconds" (which will turn the unit on, if off, for 30 seconds -- or, add 30 seconds to the remaining time displayed, if on) and "CLEAR" (or "open door") to manually stop the cycle. Much easier than manually selecting "DURATION", a numeric value, "START" (assuming you are happy with the default power level).

It's over 30 years old; the previous unit had a failed 50W power resistor to provide the "low power" mode.

Likely a typical consumer that doesn't want to understand the product or its offerings. Instead, wants to remember how to set it for each of the many uses to which it is applied.

(I have no idea how long "BEVERAGE 2" runs the maggie; but, DO know that it reheats a cold cup of tea to the temperature that I would find acceptable -- without burning my mouth on the hot liquid!)

Google prioritizes the hits that they are paid to prioritize.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics