Comment Re:suggestion for a better "google suggest" (Score 1) 321
actually no (not necessarily). i guess i should elaborate on what i meant in the original posting (i don't speak english natively, please apologize for any unclarity).
there are different scenarios how an average user might act and what he does or doesn't recognize (btw i prefer to use recognize instead of read due to [1]):
0) recognizing the top-suggested word (which is also highlighted in the text field).
1) recognizing those entries in the drop-down list that he explicitly deselects (by hitting cursor-up/-down).
2) recognizing another, particular entry at a glance as exactly the entry he likes google to search for.
now let's discuss these scenarios:
0) only very few people might not recognize the highlighted (suggested) word in the text field. so we can safely assume that an average user recognizes it and hence we can at least assume that a user who rejects the suggested word/term (by hitting 'backspace' accordingly) probably wants this word/term be excluded from his search.
1) i think it's safe to assume that explicitly deselecting a selected suggested entry in the drop-down list (by hitting cursor-up/-down) requires recognizing the currently selected suggested entry (exception: situation like in case 2, see later.) - else, deselecting it wouldn't make much sense (requires an "effort" after all ;). explicitly deselecting an entry like this can be interpreted as a negative selection (i.e. the user doesn't want to search for this entry/term.
2) in this case, the user selects the exact term/entry he wants google to search for. this implicitly means that he's probably not interested in any of the other suggested search terms (even if he didn't recognize them), as the one selected matches perfectly.
the relation to the already proposed improvements:
0) and 1) are covered by the already proposed solution variant 1 (which i called "excluding all explicitly deselected search terms"):
southwest -"southwest airlines"
2) could be covered by solution variant 2 (which i called "excluding all explicitly deselected and all not explicitly selected search terms"):
southwest -"southwest airlines" -"soulseek" -"south park" (etc.)
the elementary algorithm: so, depending on whether the user a) navigates the list using cursor-up/-down or whether he b) directly selects a certain entry, solution variant 1 (-> a) or solution variant 2 (-> b) is more appropriate respectively. solution variant 2 is also applied if a user uses both a and b.
of course we could further refine the search optimizations (i.e. distinctive solutions for cases 0 and 1, addition of intermediary or fallback solutions), ideally based on statistical results from human-computer interaction experiments (remember we're interested in average users' demands).. also note that the aforementioned explanations are only valid if the list contains distinctive/unequal entries (that's the case with "google suggest" and is true for an option list in general).
this was a general elaboration that implicitly answers your first two questions. now to your other questions:
q: what to do with "bottom-up readers"? a: doesn't really matter (in general/average) as the same algorithm can be applied (usually solution variant 2 as "google suggest" selects the first entry by default).
q: what about options 3,4,5 etc.? a: this is covered by solution variant 2 already.
i hope this clarifies most of it :) i still didn't mention all the thoughts behind it, but i think readers can find out the rest by themselves now..
[1] i'm neither a linguist nor a neurologist, but iirc, there's empirical evidence that people don't need to fully read a word to recognize/understand it. for example, most people would probably recognize "southswet ailrinse" as "southwest airlines" almost immediately without (fully) "reading" it character by character.
there are different scenarios how an average user might act and what he does or doesn't recognize (btw i prefer to use recognize instead of read due to [1]):
0) recognizing the top-suggested word (which is also highlighted in the text field).
1) recognizing those entries in the drop-down list that he explicitly deselects (by hitting cursor-up/-down).
2) recognizing another, particular entry at a glance as exactly the entry he likes google to search for.
now let's discuss these scenarios:
0) only very few people might not recognize the highlighted (suggested) word in the text field. so we can safely assume that an average user recognizes it and hence we can at least assume that a user who rejects the suggested word/term (by hitting 'backspace' accordingly) probably wants this word/term be excluded from his search.
1) i think it's safe to assume that explicitly deselecting a selected suggested entry in the drop-down list (by hitting cursor-up/-down) requires recognizing the currently selected suggested entry (exception: situation like in case 2, see later.) - else, deselecting it wouldn't make much sense (requires an "effort" after all
2) in this case, the user selects the exact term/entry he wants google to search for. this implicitly means that he's probably not interested in any of the other suggested search terms (even if he didn't recognize them), as the one selected matches perfectly.
the relation to the already proposed improvements:
0) and 1) are covered by the already proposed solution variant 1 (which i called "excluding all explicitly deselected search terms"):
southwest -"southwest airlines"
2) could be covered by solution variant 2 (which i called "excluding all explicitly deselected and all not explicitly selected search terms"):
southwest -"southwest airlines" -"soulseek" -"south park" (etc.)
the elementary algorithm: so, depending on whether the user a) navigates the list using cursor-up/-down or whether he b) directly selects a certain entry, solution variant 1 (-> a) or solution variant 2 (-> b) is more appropriate respectively. solution variant 2 is also applied if a user uses both a and b.
of course we could further refine the search optimizations (i.e. distinctive solutions for cases 0 and 1, addition of intermediary or fallback solutions), ideally based on statistical results from human-computer interaction experiments (remember we're interested in average users' demands).. also note that the aforementioned explanations are only valid if the list contains distinctive/unequal entries (that's the case with "google suggest" and is true for an option list in general).
this was a general elaboration that implicitly answers your first two questions. now to your other questions:
q: what to do with "bottom-up readers"? a: doesn't really matter (in general/average) as the same algorithm can be applied (usually solution variant 2 as "google suggest" selects the first entry by default).
q: what about options 3,4,5 etc.? a: this is covered by solution variant 2 already.
i hope this clarifies most of it
[1] i'm neither a linguist nor a neurologist, but iirc, there's empirical evidence that people don't need to fully read a word to recognize/understand it. for example, most people would probably recognize "southswet ailrinse" as "southwest airlines" almost immediately without (fully) "reading" it character by character.
