Ordinary Nothing

Looks like Yahoo! Mail finally fixed that annoying email filtering problem.

Everything seemed to work properly a few days ago, but given the rapid changes that went over there in the past I decided to give it some time. Well, some time has passed, and everything still looks fine.

I don’t know if they’ve solved what caused the problem, or just put in a proper workaround, but if I can’t see the problem I don’t have a problem.

Which means that as long as I don’t need any support from Customer Care, and don’t have any questions, they’re again my favourite webmail provider.

Main power control for personal computers come in two basic configurations/standards. AT and ATX (Well, there’s a lot more to those than the way the power supply works, but that’s the part I care about here).

AT is the way the older computers were connected. The power switch is a physical connection. You have a two-way switch, with an ON state and an OFF state. You move it to on in order to turn the computer on (surprised?), and to off in order to turn the computer off. The main advantage of the system is that you have full control of whether power is on or off. The disadvantage is that you have to press a switch in order to turn the computer off once you’re done with it. This never seemed problematical to me, but since they later developed a solution, I guess it was a problem.

ATX is the newer type. The switch is electronic. You only have a single button on the computer, which is used both to turn it on and to turn it off. This allows you to turn the computer off by software, so your computer’s operating system can provide a shut-down option that actually turns the computer off. No need to press buttons or anything tedious of the kind. The disadvantage is that it sometimes doesn’t get turned off due to a variety of reason. And then instead of flicking a switch, you have to press the stupid button for about 8 seconds until something realizes that you actually want it to turn off.
Way of the future…

We’re upgrading a system for a client. One of the changes involved is an upgrade to the computer. This is an industrial system, and we’ve ordered a new board that will fit the existing chassis and have enough extension slots for all the cards we need on it.

Before replacing the board in the chassis, we took a look at all the connections. Next to the power connections there was a jumper used to choose between ATX and AT modes. The existing button and power supply are AT. Next to the jumper there was a mode printed on the board. The jumper should be open for ATX and closed for AT. So far so good, we need to close the jumper.

Well, surprise. Instead of a jumper, there were two holes in the board where a jumper could be fitted, but no actual jumper. If we want one, we need to solder one in ourselves. And since not having a jumper means it’s effectively open, we want one.

It’s a real nice feeling, to take a brand new computer board, and start soldering stuff into it. Great way to ruin it before we can even check it. Our electronics engineer was thrilled.

And to add insult to injury, the board’s manual clearly listed that the closed option is the default one. Yea, right.

Anyway, jumper soldered and closed. AT power works fine.

A friend called me on the IM the other day, and asked for help spelling a word in English.
The friend, whose English is usually quite good, knew that the word sounded like "satel", and gave me an example case where this word would be used.
The exact example isn’t important here, just that the friend knew what the word meant.

The word wasn’t complex, it was "subtle". Didn’t took more than a second to come up with. The friend said thanks, and that while looking for various possible spellings, sticking a "B" in the middle of it did not came to mind.

And that was that.

But it got me thinking about this in a more general case.
Personally I rarely have the problem of knowing how a word sounds but not how it’s written, since most of my English comes from books, not from speaking or from TV. So if I know a word, it’s quite likely I saw it written, and could come up with something close enough for a dictionary/spell-checker to find. But if it’s a complex word, the problem may still happen. And of course a general solution could also benefit people whose vocabulary comes mostly from watching TV, speaking, or listening to radio/music.

So I decided to see if there’s a tool that can be used to locate a word, when what is known about it is how it’s supposed to sound like, and the general meaning. This is quite a lot of info.

The first attempt was just to run a dictionary search. On-line dictionaries tend to present possible spellings/corrections.

Usually I use services like Dictionary.com, since they collect definitions from a large number of dictionaries, increasing the odds of finding the right word. In this case is was not effective, the large number of sources allowed it to actually find a definition for the acronym "SATEL", so I got no spelling suggestions.

Trying to get a correction out of Google proved fruitless as well, it didn’t find any definition, and did not suggest a correction. Surprisingly it seems that Google doesn’t bother suggestion corrections to dictionary searches, only to regular searches. And only to regular searches that return few results. Satel, and variations, have tons of results, so no suggestions.
Several other dictionaries provided suggestions, but none useful.

Merriam Webster provided a list of 14 possible corrections, the 6th being "subtle". In this case it proved helpful, and could have been used.
In a general case it may not help, however, since it doesn’t provide suggestions to words it has a translation for. If the word searched is similar enough to a real word, you’re out of luck, or have to try searching for a specifically distorted word, making the chances of a successful match smaller.

There should be some sites that allow searching for words that sounds like other words. A quick search for "dictionary sound like" returned several likely (and lot of unlikely) suspects.

AnsMe provided a long list for a "sounds-like" search on "satel", but none of the results was "subtle". And #2/#1 with 90% match was "stela" ?! How does "stela" sounds like "satel" ?

RhymeZone for "Find similar sounding words" returned 0 results. It did provide a long list of "similarly spelled words", but again "Subtle" was not one of them.
Other places in this category seems to just point to, or take results from, RhymeZone.

So time to go look by meaning. Not to make it too complex, these are the first four basic ideas for meanings I had: "delicate", "gentle", "not obvious", "not blatant".

One option is to run a regular dictionary search and try to fish a result from there. In this case it might have worked, but only since I already knew what I was looking for.
Running a search for "delicate" on various dictionaries returned pages of results that had the word "subtle" in them. But they were not useful for this purpose, since it was not practical to find those without knowing the desired word. It takes a lot of time and effort to read many definitions, and go word by word to see if something seems similar. Doable, but not practical.

Another option is to run a search for synonyms of the first two words, and antonyms of the last two. This has the advantage that the results would usually be word-lists, so are easier to search for one sounding like what we’re looking for.

Giving RhymeZone another chance: "delicate" 23 synonyms including "subtle", "gentle" 25 synonyms without "subtle", "obvious" 3 antonyms without "subtle", "blatant" no antonyms. So there was a result, but still hard to fish. Requires going through a lot of false result to find it.

WordNet: This is a more complex and powerful tool, in that it allows to pick specific meanings of the word to search the synonyms/acronyms for, though it doesn’t require it. This of course takes the time to read the possible meanings, but allows by this to eliminate irrelevant words. It also provide examples for the usage of each of the returned words, so it’s easier to get a sense of their proper context, or how they will sound in a sentence. Here too only "delicate" returned "subtle" at the end of the search. And while it was included in some of the definitions of possible meanings for "delicate", no meaning was an exact match by itself. It might have helped very well, or it might have done quite badly.

And there’s one more, relatively new, option. I recalled that a while ago I noticed an announcement by OneLook that they had a reverse dictionary. For this purpose a reverse dictionary should work quite like a synonym search, only will potentially allow use of words which are not exact synonyms/antonyms but may be found as part of a description. This makes it easier to locate words for which you don’t have an exact single-word meaning, and allow more flexibility on the words used. But it may also return a lot more false results.

I decided it’s worth a short to try. And discovered that they have another feature, which is excellent for the sort of search I need here. It’s possible to search for partial matches on words by letters, and combine this with the reverse dictionary search.

The obvious nearly catch-all way to represent the "satel" sounds-like is s*t*l*. It may be a mistake, it’s possible to spell words otherwise, but not likely. If it wouldn’t have worked, maybe a search for c instead of s, or putting another wild-card character at the beginning, might have been warranted. But mostly I tried to go from the basis of having a word sounding like "satel", so probably the changes in spelling will be extra vowels, double letters, silent letters, and such. Something beginning with an "S", followed someplace by a "T", and followed someplace else by an "L", followed by whattever, seemed good.

With this pattern of searched words, I tried again my four basic meanings, even though a reverse dictionary actually allows to try for more complex meanings. "delicate" returned 7 results, the first of which was "subtle". "gentle" returned 2 results without "subtle". "not obvious" returned a huge number of results, with the claim that they are sorted by relatedness, and "subtle" was the first one. "not blatant" also returned a lot of results, but again "subtle" was the first one.

Overall I think OneLook receives best marks for this. The reason, I think, is that from all the common tools available it’s the only one that allows to automatically search for both the meaning and the way the word sounds like. The other services only allowed to automatically search for one, and then required time consuming effort to manually go over the results.

[Update, the last: They finally fixed it!]

[update: Well, They've done some more. And I'm practically livid.]
[update2: There seems to be some progress on the technical side, although I can't grasp the logic of it]
[update3+4: Some progress. Not done yet, but the filtering problem seems practically solved for now]
[update 5: Back to square one. Everything is as it was before they've started working on the problem, and I can't get anything sensible out of customer support]

There has been some progress changes regarding the problem I encountered with Yahoo! Mail’s bulk mail filtering and my futile attempts to get help through their official tech support system.

A certain
Brad Garlinghouse (VP, Communications Products) from Yahoo! made a post on the Yahoo! Search Blog regarding changes to Yahoo! Mail, and asked for general comments and suggestions.
So, being happy to finally have access to a real person involved with Yahoo! Mail, I talked about the problem, with links to my more detailed posts here.

And, not too surprisingly, some people at Yahoo! read it, and they tried to solve the problem.

I started sending myself test messages to see if there’s any change, and after a short while I noticed a rather big one. When I sent myself a problematical message, I was presented with a CAPTCHA challenge. One of those annoying pictures with distorted letters and numbers, used to make sure I’m a real person sending the message, and not some bot.
It came with an explanation that this is done to prevent outgoing spam out of Yahoo! accounts.

Which is overall fair enough if the message is highly suspected as spam. Only it probably isn’t since if I sent a similar message in from an outside account it would not get to the Bulk folder but to the Inbox one.
This is also not directly related to the fact that even if a message is suspected as spam, it should still get to the Inbox if the sender is in the Address Book.
But the important thing was that after passing the challenge, the message arrived to my inbox.

If I sent the same message to another Yahoo! address, I got the challenge. If I sent it to a non-Yahoo! address, I didn’t get it. It seemed a bit strange if the reason is suspected massive spamming (And never mind why is sending a geocities link in a message have to be considered spam). So I posted another comment on their blog post.
And they fixed the apparent problem. Later on when I sent the message outside of Yahoo I also got the challenge.

Again, sending such a message in from an outside non-Yahoo account always got it to the Inbox, so I’m still not sure the problem is spam detection per-se.

But the problem was officially solved. Even if it’s a case of curing the disease while killing the patient. After all, the reason I was sending myself emails with links is as a short and easy way to place reminders to places I’d like to look at again soon. But passing the CAPTCHA challenge made it long and cumbersome, so the net effect would be to just cause me not to use Yahoo! Mail for these messages but another email service. Which is a pity since overall I really do like Yahoo! Mail.
But the problem was solved, technically. If I did send such a message, it arrived to the inbox. I did suspect that the actual cause was not fixed, and this is just some ugly workaround, but there was no possible way for me to reproduce the problem, so that was it.

Until today.
Originally, beside having the mail problem, I also had a larger problem with the handling of my report by the support / customer-care team.
As I logged in this morning to check emails, I noticed a message from Clarence in Yahoo! Mail Customer Care. It was a followup to my previous correspondence with them on this problem. And by followup I don’t mean that they started a new thread about it, but that it was a reply to one of my messages to them. Not to the last one, mind you. In fact it was a reply to the exact one where I told them I’m not willing for them to log into my own account to check the problem.
And what did they want this time?
Well:

We would like to follow up with you regarding your recent inquiry to Yahoo! Customer Care. 

We understand you were receiving messages to your bulk mail folder.  We attempted to duplicate the issue by following the steps you outlined below, yet were unable to duplicate it.  Are you still experiencing problems of this sort?

If the problem continues, please reply to this email. 

For us to look into the problem you have encountered, it will be necessary for Yahoo! staff to enter your account and conduct some tests.
Please reply to this message, giving Yahoo! permission to enter your Yahoo! Mail account and take those steps necessary to pinpoint the cause of this problem and explore possible solutions.

We appreciate your assistance in troubleshooting this issue.

Notice that the "For us to look into…" paragraph is identical to the one they used the last time. This is particularly amusing considering that this is, as I wrote above, in direct reply to my message saying I’m not allowing them, which was in reply to them asking. And it’s all quoted in this message…

But in any case I was glad that I got a response. It means someone may have also noticed a problems in this front, and there may actually be someone there to talk to (Clarence?).

I was not surprised they were not able to reproduce the problem, since there were changes. So I decided I’ll reply stating the changes I know made, and thanking them for following up anyway. But before I sent the reply it seemed prudent to make another check and see if something changed.

Something did.
I sent the same message to myself, and was not presented with the CAPTCHA challenge. My initial reaction? Great, maybe the finally solved it properly and so removed the challenge that served as temporary work-around. I fired off one more message, and went to the Inbox to see them. The messages were not in the Inbox, they were in the Bulk mail folder.

Back to square one. Several changes along the way, and now the problem is just like it was before…

Which leaves me the possibility that the Customer Care people will be relevant and professional this time. Maybe Yahoo did something on that front, at least.
I sent them a message explaining the changes, and that maybe they need to try again now in order to reproduce the problem. (In a perfect world they should also be in synch with the development team, but…).
I also stated again that I don’t want them in my account. In case they will have a problem again, I did offer to create a new dummy account, verify that the problem occurs there as well (Shouldn’t be any different, it happens in several real accounts of friends), and give them full access to that one.
And I explained again how to reproduce the problem, in small and easy steps.

If any reader here wants to try, it goes about like this:

1. Add yourself to your own address book. (This is to illustrate that the message goes to the Bulk folder even for AB contacts)
2. Send yourself a plain text message with:

• Subject: Test

(Can be pretty much anything as far as I noticed, but this one is verified)

• Body: http://www.geocities.com/whatever

(You can as far as I noticed replace the "whatever" word with anything you want, as long as something is there, erasing it won’t work. adding further sub-directories, and a final  page, will still cause the problem.)

Then just sit back and watch it get delivered to your Bulk folder. Despite it being a message from you to yourself, and despite the sender address being in the address book.

Now I wait. I’ll keep updating this post if there are changes to the actual problem, or if there is some progress (or an active lack of of one) from the Customer Care people.

Update:
Instant messengers are wonderful. Especially those that come with email providers like Yahoo. Because when you get a new message, the IM let’s you know.
For example: When people from Yahoo! Customer Support log into my account DESPITE BEING SPECIFICALLY TOLD NOT TO, and start sending test messages, I get to bloody see it. Amazing, isn’t it?
Even if the messages are deleted immediately afterward, I know they were there.
Better still, I refreshed the Inbox page fast, and actually saw a message labeled as being sent from me, only from the wrong timezone and IP addresses. A message which was not subsequently deleted and is currently still there, I assume because I opened and read it (since the previous ones, that I didn’t see on the webmail interface, are not there now).

Where they told not to clear enough? Let’s see…
This is from the message I sent them, as specified above. Everything quoted here is quoted in the message body by the replies. I stripped irrelevant lines, but everything here is in the message, no additions. Notice that as this is an email, the oldest bits are in the bottom, and the newest at the top.

Let me start again by saying that I do NOT want you
logging into my account.
If you can’t see anything on your accounts, I’m
perfectly willing to open a new dummy account, make
sure the problem occurs there as well, and give that
to you. Is that acceptable?

Let me know whether you manage to reproduce it now or
not, and if you want me to create and hand over a demo
account.

— Yahoo! Mail <mail@yahoo-inc.com> wrote:

>
>
> If the problem continues, please reply to this
> email. 
>
> For us to look into the problem you have
> encountered, it will be
> necessary for Yahoo! staff to enter your account and
> conduct some tests.
> Please reply to this message, giving Yahoo!
> permission to enter your
> Yahoo! Mail account and take those steps necessary
> to pinpoint the cause
> of this problem and explore possible solutions.
>
>
> > Original Message Follows:
> > ————————-
> >
> > NO. The problem is general, and NOT just with my
> > account, so there is NO justification for you to
> see
> > any of my mail or enter my account.
> >
> > To make it clear: You do not have my permission to
> > enter my account.
> >
> >
> > — Yahoo! Mail <mail@cc.yahoo-inc.com> wrote:
> >
> > > For us to look into the problem you have
> > > encountered, it will be
> > > necessary for Yahoo! staff to enter your account
> > and
> > > conduct some tests.
> > > Please reply to this message, giving Yahoo!
> > > permission to enter your
> > > Yahoo! Mail account and take those steps
> necessary
> > > to pinpoint the cause
> > > of this problem and explore possible solutions.

Did the fact that I said "I do NOT want you logging into my account" mislead them? Could anyone possibly understand this phrase to mean that I don’t particularly want it, but won’t mind if they do?!

I know that technically they can log into all mailboxes/address-books/notepads/whatever and do anything they want with them, but it’s not accepted behaviour, and not something that their privacy policy allows. This is why they asked. But darn in, You must accept NO for an answer. I’m pretty sure this was decided by courts long ago…

True, one shouldn’t keep anything unencrypted that one doesn’t want the whole world to have access to, but there is still a fake feeling of privacy for things which can become public but shouldn’t. And knowing that support people are running around my mail and contacts is not a particularly nice feeling.
Even if I trust that they are professional enough not to touch anything not needed for the solution of the problem.

Do I trust that they are professional enough, with this past record? Not quite…
Heck, this one message that I did catch, that got into the Inbox? It had "test" in the subject, and "test" in the body. Not the URIs which cause the actual problem. Despite repeated explanations.
No wonder they were not capable of reproducing the problem. They either ignored the explanations, or are bloody morons (And yes, I can use mildly foul language when I’m in the mood I’m currently at).

I ran some more tests myself. The first one got to the Inbox and not the bulk folder. But for some reason it was not marked with the Rolodex icon (meaning that despite being sent from the same address, it’s not recognized as from someone in the address book). The two reached the Bulk folder as before. The following one took about 10-15 minutes to arrive, but arrived at the Bulk folder as well.
Problem not solved.

Update2:
I sent some more test messages to see where they’ll get to. Got some peculiar results, although overall I think it’s an improvement. Three different URIs in the message body caused three different results:

1. http://wwww.geocities.com/testing/abc.htm  -> Message arrived to the Inbox as it should.
2. http://www.geocities.com/whatever -> Message arrived to the Inbox, but there is no Rolodex icon next to it. Meaning that the Yahoo! Mail interface decided the sender is not in the Address Book. The sender of course is identical to the one in case 1, where the Rolodex icon does appear, and is in the address book. In this case there is no "View Contact Details" link next to the sender when reading the message. There is one in case 1. The address itself is displayed properly.
3. http://www.geocities.com/bli -> Arrives at the Bulk folder. So the problem is not yet solved.

I sent several messages with each link, the results where exactly the same for each link. Separating case 1 from the other two seems straightforward (wwww instead of www), but I can see no conceptual difference between the 2 and 3 URIs. Trying to keep the same structure as case 2 and 3, replacing the work "whatever" and "bli" with other words, the messages arrived to the Bulk folder like in case 3.
Does it mean the word "whatever" got a special case ?! Or is there a rule that I just can’t see?

… After writing this I tried again with another word, and got the CAPTCHA test again… So either the handling is even more complex and obscure, or I just caught them smack down in the middle of work, which can account for odd behaviour.
So I’ll stop checking for now, and check again in after several hours.
The current CAPTCHA test is very barbaric, though… I apparently failed the first attempt (They didn’t bother getting rid of letters that look the same in both lowercase and uppercase), and instead of getting to try again I was told that I need to pass the test in order to send messages, so the message was not sent and was not saved. No options to go back.

Update3:
Shortly after the previous post the system seemed to revert to the original state, just as it was at the beginning, with the mail delivery problem.
Nothing much seemed to happen for a few days, I didn’t hear anything more from the Customer Care team, and sporadic test emails behaved in the same fashion.

Today I emailed another message to Yahoo Mail’s Customer Care, asking what is being done regarding the problem and if there’s any progress (There was a bit more there, like a complaint that they logged into my account, and asking if they need more explanations, but that’s the gist).

I don’t know if it’s related to my prompting, or just a coincidence of timing, but right now I sent some test messages of the same problematical patters, and they arrived into my inbox. Well, two out of three did, the third is in transit. I assume it would as well, if not I’ll update here again (Update4: Arrived into the inbox, same as the rest).

So why am I saying that the problem isn’t entirely solved? Since those messages (from me to me, with a geocities link in the body) are showing in my Inbox without the rolodex image (the one indicating that the sender is in the Address Book). But messages from me to me without a geocities link show in the Inbox with the rolodex image, as they should.

This means that whatever was done to solve the problem, was not geared toward correcting it, but toward making it appear as though it’s not there. The crucial symptom is gone, my messages do not get into the Bulk folder. But something is still wrong in the logic that classifies them differently than other messages.

Update5:
For the last several days the messages are back to being received into the Bulk folder. Everything seems to behave just like at the beginning, before anyone at Yahoo! started to deal with the problem.
Obviously they know something is there, or they wouldn’t have made the repeated changes I saw, but they can’t seem to actually solve it.
Personally, whatever else goes behind the scenes over there, I’d expect that a simple match against the Address Book should be enough, but that doesn’t seem to be the case.

As for their Customer Care, I got a response to my message, but it only address the CAPTCHA challenge. A certain Jane explains that the challenge is meant to prevent automated spam tools (Yes, I’m certainly an automated spam tool, sending a single-digit number of messages per day at all, and a smaller single-digit number of messages to myself). She also informs me that they use "a variety of techniques" in order to decide when to present the challenge. And concludes by explaining to me how to pass the challenge (‘Cause obviously I can’t get by myself the idea of writing back the word I’m seeing).

Since the challenge is no longer presented to me on every sent messages, this is moot anyway.

I sent another reply explaining that the problem is not the challenge, but rather the messages getting to the Bulk folder, and that she should read further back in the message if it wasn’t clear. This is why you should let one person deal with a problem start-to-finish, and not pass each reply/iteration to a different person who doesn’t know what it’s about.

At this point I think I’ll give up on them.
The technical problem is obviously too complicated for Yahoo! to solve (Can’t match the sender with the AB entry, or match the sender with the recipient, or realize what’s the problem with geocities).
And the Customer Support ability to grasp simple English, or follow a conversation thread, have obviously not improved any.

It’s a pity that overall Yahoo! Mail is still much better than the competition. But apart from this problem, and as long as I don’t need anything from their support, it’s really good. I don’t quite get how they can deploy a generally very good service while using seemingly incompetent people, but they do.

More updates will come only if there’s something serious to report.

Either Ask Jeeves have some bored new employees, or they officially have some serious concerns about they way people see them (Nice, but not too good in relevancy) and believe that their blog indexing abilities are somewhat lacking.

During the last day this not particularly popular blog of mine has gotten two hits from the ask.com domain. Both from a Bloglines subscriber running a search (a subscribed, repeated search. Not a one time search) there (and not in Ask Jeeves). Says something about how well they believe they index blogs.

One search was for "Ask Jeeves" and one for "Jeeves".
Since I have one entry (Hey, that’s two now) that mentioned Ask Jeeves, it was shown in both their search results. And despite it being very obviously not relevant (but then again, we did say that Jeeves had a relevancy problems with searches, no? Maybe it starts with the real humans and propagate to the engine from there?) they clicked through to read the full post.
And yes, the excerpt is enough to ascertain that the post does not talk about the search engine, but only mentioned in passing that it’s about the question that got into my referrer log.

Is this a good thing, that they show an interest about what people think about them? Or is that an indication that there are problems over there and they’re desperate for something to help?
And what should it mean that the team of a large search engine uses a different search engine to constantly monitor themselves? Is it OK since they don’t index blogs purposefully? Is it an indication that they just now decided blogs are important and are in fact working on it?

Or was it all just a bored employee that wants to know what people think about their company, and picked what they thought to be the best tool for the job? Still says something…

I find it amusing, in any case. Maybe Jeeves won’t.
At least this post is somewhat relevant, if they get here.

What a truly appalling idea. Sending ads to people’s cellphones when they come near a store, and charging a fee if the person (or at least their phone) enters the store.

Personally, receiving ads on my cellular phone would actually strongly discourage me from buying anything from the ad sponsor. I might, however, go in and out of the store repeatedly, just to make them pay (pun intended).

More than that, this idea has some serious technical problems which I don’t quite see as solvable.

If people receive the ad when they come close to a store, then any person intending in advance to shop there will receive it too. And when they get into the store, the store must then pay for a supposed ad-induced entry… Unlike a web click-through, there’s no good way to separate people that entered because of the ad from people that entered regardless (or despite) of it.

Just brilliant.

The comparison to web advertising would be more like placing an ad at the home page of a site, promoting that selfsame site, and having the owners pay for every viewer of the ad.
Or more realistically (if I must), it’s like tracking who saw an ad, and making a site owner pay when they go there, even if not by a click. Hey, since large ad provider actually track ad viewing today, they could implement it right now. Whenever you see an ad, they can check their cookie to see if you saw a previous ad for the site you’re browsing now. And if so, make the site owner pay. It’s the exact same model. Yet somehow I don’t see people paying for it… So why should it work better with brick-and-mortar?
Paying for anyone who shopped at your place, and who have also seen an ad, is not a good idea. The connection is too tenuous.

Not to mention, since some people are like me in the great love for being spammed with ads on their phone, would the ad service also offer a refund for people getting the ad and not going into the store? After all, the ad sponsor could claim that it was a potential customer who was lost due to the ad… The connection between seeing the ad and not going in, and seeing the ad and going in, is of about the same strength. If you accept one, you should accept the other, no?
Hmmm… This actually could catch up on the web. Anyone buying ads would love to get paid for ads that people saw without ever going into their site afterward. If an ad provider offered this, I’d consider buying ads from them myself…

HDTV on cellular phones ?!

Yes, with current huge screen sizes on cellphone, surely the higher resolution of HDTV is the main reason the technology still didn’t catch up. The displays themselves are already extremely high quality, and regular TV movies would just look incredibly granular. No doubt.

How high a resolution do you need on 1″-4″ screens for crying out loud?! Seeing regular TV shows on a computer screen (Only old and public domain shows, of course), if I stretch squeeze the display small enough, they look perfectly fine. Higher resolution lets you see better on larger screens, but is totally wasted on tiny ones.



Nuts.



Hat tip to Mike on Techdirt.

The Bank of Israel has published a new press release (Hebrew only I’m afraid), announcing the completion of a project to backup it’s data systems in case of an emergency.

They built equivalent computer systems to their main site. On two different remote sites, one against their central computer, and a second against their networking and servers. I assume that the central server would be actual banking and monetary data, and the network servers are all the other documents, mail, files, and so on.



They talk about continuous data mirroring to the remote sites. So if they have continuous data mirroring, and identical/compatible computers at both ends, they probably mean they run Clustering.

Which is good. Properly arranged Clustering allows very rapid recovery in case something happens and a site goes down.

In this case, the rate of recovery seems to be 6 hours for the critical systems, and 3 days for the non-critical ones.

Hmmm… Is it just me, or is that way way way too much for this? If they just made remote backups of files and databases, I could understand the time it takes to install everything and get it online. But they have full high availability clustering. Or at least they should, since otherwise making “continuous data mirroring” to “equivalent computer arrays” (rough literal translation) is just a total and colossal waste of money, and bad design. Assuming of course that everything in the press release is accurate, and not phrased specifically to give wrong impressions.



Oh, well. So the recovery times suck. It’s not as if this is the central bank of the country, that all other banks depends on, and that is in charge of all the monetary policy of the country… Oh, wait, it is.



Well, that’s the way it is. Me bitching about it won’t help much. At least they did all the project remarkably fast. It only took them 10 months… To purchase and install computers and EMC data storage devices, network the sites, buy and install proper clustering/backup software. I can see how it would take a month. Two, this being the public sector. More than that? Working with two outside contractors, which should be able to work fast once they get the purchase order? I’m not impressed. At all.



The project is reported to have been finished on March 2004. After an “intense” 10 months efforts. What’s the hurry, you ask? Well, it seems that this is the result of lessons learned from September 11th 2001. Let me see, finished 3/2004, means it started in 5/2003. So getting “one of the most important lessons from…” took only… err… slightly more than a year and a half? Cool. Good to know we have some fast thinkers there.

That is, assuming it has anything to do with it. Which I’d suspect not. But since everyone in the US now blames everything they do on September 11th, they must have felt they can too. I mean, it obvious you need a good excuse in order to create backups of critical infrastructure. Nobody does it otherwise. That is, if you expect people to ask why you decided to do it all out of a sudden, instead of expecting people to ask why the heck didn’t you do it year and years ago? The technology has been available from a lot earlier. I know it was. I used some of it, and it was still the 199Xs then.



They also claim they ran an experiment, simulating a disaster, and switching to work on the remote systems. They don’t quite specify which systems, but the sense is that it wasn’t an all-out case. This sense being supported by the fact that the recovery times I stated here are not measured, but rather estimated based on “mapping all the systems in the Bank of Israel in an attempt to determine the time duration requires for their recovery in case of a disaster, and the scope of the data that needs to be restored”. So they don’t know how long it would really take. And apparently are not restoring everything, even after those 6 hours or 3 days. Good to know. That’s exactly what experiments are for, people. Run a full one, and get an actual and reliable idea about recovery! Do they even know that recovery works?!



Well, at least now I can sleep safe at night.

The nice people at The Register (Well, it’s very hard to catch them actually being nice, but in this specific case either they are, or their sarcasm is much subtler than usual) have published an article covering a new social-networking thingy called Yelp!.

The idea behind Yelp! seems to be that it lets you ask your friends for recommendations about places. And (This is the social-networking part, I suppose) then for them to go on asking their friends for you. All with less work and hassle, at least according to Yelp!.

Let’s look at the workflow, by the nice how-it-works from Yelp!’s site. Step one:

Ask Your Friends

When you want to find good local businesses you ask the people you trust. Yelp! makes that easy.

To get started, just enter what you’re looking for, together with your friends’ email addresses.

Quite amazing so far. I only need to write what I’m looking for, and my friends’ email addresses. Let’s compare to the other option, sending an email (since “Friends” here are more “people that won’t be too bothered by a query” than actually friends. Otherwise I might have been tempted to use the phone).



If I want to ask my friends something by email, I’ll need to enter all their email addresses, or pick them from the address-book.

Through Yelp! it’s pretty much the same things, except I don’t see an address book. Maybe once you register you get one with all the addresses you previously used.

If I want to ask in my email about something, I have to laboriously write it down in free text, having to come up with all that grammar and syntax needed to make my question sensible. I can’t just write someone with a line like “restaurant Foo city can’t cook” and have them understand. I’ll need to make a whole paragraph like “Hi. Listen, you know I can’t cook, and I’m getting hungry. So I was wondering if perhaps you know of a good restaurant someplace in the city of Foo? Thanks a bundle. Oh, and if you don’t know of any, mind asking around for me?“. That’s work, work, work. Very difficult. (I can BTW. Cook, I mean. And very well too. This is just a theoretical example. OK?)

With Yelp! on the other hand, you just fill in three boxes on a form, one for what you’re looking for, one for the city, and one for extra info. Then they can spare you all that sentence-building. No reason for you to actually want or need anything more when asking a friend for a recommendation.

OK, onward to step two:

Friends Make Recommendations

Friends will receive your request by email. Then, they recommend a business using Yelp! (sign up is not required).

If they can’t help, they can forward your request to one of their friends.

My friends will receive the request by email, and can recommend a business. If they don’t have a clue, they can forward it onward.

Differences? If I send them an email, my friends have the burden of replying and writing a whole reply, and sending it back to me.

If on the other hand I use Yelp!, instead of the onerous task of pressing the reply button, they can easily click a Yelp! link. And put the reply there. Better yet, my dear friends could then fill in a form with all the required details in the proper fields, not being forced to do this whole free-text routine themselves. Heck, using Yelp! they don’t even have to register! (Oh, wait, same for replying to an email…)

Seems obvious I’ll get more replies if replying requires less work. All the hurdles of writing a line in an email is sure to discourage some people that would be perfectly happy to go to a website and fill up a proper form with fields and everything.

Stating they can reply to one of their friends is also quite wonderful. Prevents all that avalanche of chain letters I usually see when friends forward my questions to hundreds of people without any semblance of control… (OK, I’m probably just being petty here, I’m sure they didn’t really mean one. But I can’t be certain, can I?)

What’s next?

Return to Yelp!

When your friends respond, you get an email. Just return to Yelp! to see their recommendations.

Each recommendation includes contact details for the business, a map and your friend’s opinion.

If there are any replies, I get an email. Fine.

If I was just emailing my friends, this email would contain the address and details. And that would be that. Is it really fair to finish the fun so quickly? What will I do with the rest of my free time?!

Since I’m using Yelp!, however, there’s more for me to do. I can go to their website and see the reply, that my friend sent me, there. With whatever pretty colors they can put in their design, and not the dreary email text. Much better.

Ah, but I’m ignoring the really useful part. I’ll get:

• My friend’s opinion. What a novel idea to send me that. How did they think of that?


• Full contact details for the business. That’s good. If I want to call them and ask something I don’t need to open the phone book. And I get the exact business name instead of some sounds-like that my friend would have sent me from memory on the email.


• Map. So I know exactly what it is and how to get there, instead of opening one myself (or going to a mapping site and deciphering the location from the vague “one turn after the big tree” sort of things that my friend would have probably sent me on the email).

Just a slight problem with the last two items here. You might have noticed. The details have to come from someplace. Now, I won’t claim that there are only two options, but the number is limited:

1. The friend enters all the contact details, and exact address for their mapping service to find. This does spare me the work of finding these out. From my friend’s side, however, it forces him/her to do all the hard work. I’m getting a small favour, and require more instead of saying thanks.


2. Yelp! connect to external comprehensive maps and contact details providers. The friend use these and partial info to find the exact data. Potentially this can require less work from my friend. It’s still work that I’m the one who is supposed to be doing, since I’m the interested party. And I likely won’t need all the info, so it’s needless work. Incomplete date in their database for some possible businesses is of course not an option worth mentioning. Oops, I just did.

And that’s about it. Even Yelp! agree:

That’s It!

You’ve found what you were looking for. Best of all, recommendations are saved so you can share them with friends.

With Yelp! you’ll always be able to find the best local businesses.

Asking my friends about the best local businesses can sure help me find them. And yes, Yelp! is a way to do that. I’m not so sure about that always bit, and I’m not so sure Yelp! will help me more than plain old email or phone (IM too. I can ask my friends about local businesses on IM as well).

I’m also not so sure it won’t help me less, overall and all things considered.



Sharing is probably the main advantage, social-networking and all that, you know. They are not too detailed (read: doesn’t say anything last time I checked) on their website about the sharing part, and only slightly more about the questions forwarding part. So I’ll just go on assuming. They’re probably more imaginative than me over there, so I could be wrong, but then again I could be very wrong.

Suppose I ran a search about a business in a city. Suppose I even got an answer. Suppose the system even let me rank the answer (or answers) I get. Now a friend of mine runs the same search. Will it help him? Not likely, he may be using slightly different words, and that’s that. Just read their about page and you’ll see what I mean. Their CEO will run a search for cheap eats (at least you know they’re not spending money on salaries) and their CTO for high-end Mexican cuisine (or maybe they do, but on the IT people, as they should). If I were friends with them, this won’t help me at all, if for example I just want a “nice restaurant”.

And reliability is a major problem. Sure, friends won’t lie to me (Though the kind of “friends” that won’t find it strange I’m using some web service to talk to them instead of a phone or email, those might). But taste varies. A lot. Some questions I’ll ask certain friends just to know where to stay away from. On the first step, that’s not a problem, I just won’t list them. But on automatic searches to deeper levels of connections? The system can’t know who I trust and who is compatible with me on a new search.

Maybe it asks for any forwarding level. But in this case, it doesn’t help on networking. Choosing manually which people to forward a query to… is exactly what people would do if they got a regular email with the query.

One can only hope that even if it picks old recommendations automatically, it doesn’t automatically send queries to people. That could get very annoying very fast. Especially as you get further away from people you know.

And there is the privacy issue. Some people (myself included) don’t like their friends giving their name and address (email or other) to some company and websites. Doesn’t even matter if you agree, as long as you notice that some of your friends may. So you can’t use those friends when searching through Yelp! or they will send you back a scathing and not too helpful reply. They may know some good places too, though. So you have to send them an email. Well, while you’re sending emails anyway… You can just add some more names to the BCC line and be done with it.



Plus, I don’t like the name. Before today, if someone asked me to name five web sites that put an exclamation mark in their name, I would have very quickly answered “Yahoo!”. Then it would have taken me a lot longer to pick up another. And I probably would have broken down pretty quickly and picked “Yahoo! Mail” and the likes.

And now we have another highly self excited service, with an exclamation mark. And more, the name starts with a Y. Might be nice if they have a gigantic success. People may decide to shorten the name to Y!… Then the lawyers get it, and the fun really start.



Which reminds me, on an unrelated side note, some search engine really have to start indexing punctuation. I can’t run a search to find all places with a “!” character in the title!

This AdWord is in Hebrew. Since that doesn’t quite make it a world-wide seller, and the site I was seeing this on, The Register, did not have anything else relating to Hebrew or Israel on the page, I assume it means the ad was selected by IP geolocating and not just by matching words…



Regardless of this, As you might have noticed, it didn’t come out particularly readable (Yes, there are other characters in Hebrew beside “?“. Honestly). Not that I’m surprised or anything. But even when I tried to manually change the page encoding to various Hebrew and Unicode formats, the ad retained its original appearance (Well, I shouldn’t expect these changes to propagate like that, should I ?).

Which made me wonder why bother? Someone paid good money to place that ad. Newsflash : Unintelligible text doesn’t sell, fellas. No one will click on the link out of idle curiosity.



Yes, the text is there. Yes, it’s just an encoding problem. But I need to work and waste time in order to read it, and quite frankly ads are meant for people that don’t already feel strongly enough about you to waste time and effort.



Hmm… Maybe it looks alright under some defaults with some versions of IE? I hope it at least looked alright on the computer of the marketing guy (or gal, I don’t discriminate. Idiotism is a cross-gender issue) that wrote the ad and bought it. I wonder how tweaked their system was for this.



English can be read everywhere in every browser on every Internet connected computer (I’m ignoring whatever modifications China may require, of course). Even if you look for a targeted audience, do stick to English. I probably won’t click on your ad anyway, but at least it will be because I know it doesn’t interest me, not because I’m not curious enough to waste time finding out if it does…

I followed a link to a news story on some AsiaOne site.

At the side of the page there was a very large flash ad, published by AsiaOne, for… Google AdWords.



Oh, I admit they couldn’t replace it with an AdWords ad, since they want a specific ad on their own site, and that’s not what AdWords does.

Still, I find the whole concept of placing one ad in one medium, in order to publish another ad service using a different medium, to be very amusing.

I mean, OK, you can’t use AdWords to publish AdWords on your site, but you can at least place a text ad and not a flash one. You are after all trying to convince viewers that text ads work, no?



Needless to say, their website contain many other adds. All I saw were either by themselves, or by DoubleClick. Not an AdWord in sight on any page.

I wonder if it’s because they think it’s good enough for their viewers but not for them, or if it’s because Google didn’t want them…

Another serious goof on the Hotmail / MSN site. They’re getting good at that lately

They are posting links to their own shopping site, with a nice Shop pink! (And raise breast cancer awareness) link. That’s good.

But on some pages, the area they reserve for their sidebar with the links is a bit smaller. OK, so it does not seem smaller, and has the same width, but the pictures may be larger resulting in smaller area of text per item. To solve that they have those wonderful automated scripts that just clip the end of long sentences and replace them with ellipses. Saves space. Problem is, if the object the sentence is right at the end, it might get a bit… lost.

So on the main login page, everything is fine, when seeing various other hotmail pages everything is fine. When sending a message however, that nice clipping thingy starts to work…



End result? MSN encourages everyone to help raise breast cancer



Hmmm… Should increase their sales of related products and medicines. But I’m not sure overall it’s such a good idea for everyone. Don’t do anything MSN tells you to without thinking about it first.



Better be aware of things before you start to raise them, that’s what I always say. Mabye if they heard that, and had exhibited a tad more awareness, they wouldn’t raise anyone’s hackles…

I’m apparently a very bad worker. I saw my work hours for the last month and it turns out I had one very unproductive day in which I managed to do -11 (yes, that’s minus eleven) hours of work.



I don’t get paid by the hour, but my boss likes to feel on top of things, and prefer that we’d sign in and out of the office each day. The results are not that accurate, since I do tend to forget and later on enter approximate times (not a biggie for either of us, since money is not directly involved), but they’re generally representative.

Then all the start and end times are typed by our company’s secretary manager’s assistant into an Excel sheet, which has pre-made formulas to calculate total work time. (And just to emphasize, I did not design the thing, or was involved in it’s making in any way whatsoever. It predates me)



So how does one get such a wonderfully effective workday as I did, you ask? Simple. When entering the data to the Excel page, the start hour (approx 11:00) was dutifully copied to the Excel page. The end hour was not (Or was, but was later accidentally deleted by someone, NM). Hmmm… An empty hour field… It needs a default value… How about 00:00 ? Yes! Great!

How are the working hours calculated? By the most basic and foolproof system possible.

WorkHours = EndTime – StartTime

So if I’d come in at 11:05 at left at 21:35, I’d have 10:30 total hours of work. And if I’d come in at 11:05, and left at… say… 00:00… I’d have -11:05 hours of work. Which of course tally up into the monthly work hours, those being logically enough a simple sum of the daily work hours.

Thank goodness I don’t get pay by the hour! that could have been around two uncompensated work days , or even a bit more (The day in which I got the negative hour count, and another 11 actual hours of work to even in out).



Naturally the spreadsheet page contained no sanity checks. What errors could possibly creep up on such a simple and straightforward calculation? It’s foolproof I tell you, foolproof! And designed by someone that forgot there’s always a bigger fool.

Not that it’s that simple, mind you. What would you suggest, setting the default leaving hour to 24:00 ? We rarely leave the office that late, so the costs may be incurred by the wrong side…

Oh, wait, it’s actually quite fortunate we rarely leave the office after midnight. I mean, that would legitimately and correctly place us doing things like coming in at 9:00 and leaving at 3:00, for a total net time of -6 hours for the day… Not that bad considering we could cover most of our losses the next day and even get some actual hours of work out of it…



Then it gets funnier (sadder?). See, the same ingeniously design spreadsheet also calculates overtime hours. (No, I don’t get paid by hours, so I certainly don’t get overtime hours. Is that a reason not to bother calculating them? You think?). The formula for that is also quite simple:

OverTime = WorkHours – 9.1

Yep, 9.1 was arbitrarily decided to be the daily work hours. Can’t see why I should mind either way, considering it doesn’t effect my pay in any way, so why not?

Now can someone guess what the total productivity for this day has been for me? Let’s see:
-11 – 9.1 = -20.1

Wow! What a day, eh? I must really be something special.



Makes me so sorry I don’t come to work as late as 15:00. Now that have been really something. Not everyone can work less than minus 24 hours a day…

The login page to Hotmail has, beside the required address and password fields, various links to other content from other MSN services. Usually written as supposedly catchy phrases some PR flunky came up with to draw click-throughs.

And when you put a commercial, even if it’s a commercial to yourself, on a page that lots of people go to, you usually try to get them right. I bet they use a spell checker all the time. Making a spelling error would look really bad, after all.

The problem with spell checkers is that they find misspelled words (Well, that’s also the point of spell checkers, true. Maybe the problem is that sometime people forget that this is all what a spell checker does). They do not find wrong words which are properly spelled.



For example suppose you had an article or, a bunch of articles, like Hotmail had today, talking about various real-estate issues. What sort of text would you use to link to it? Maybe if you work for MSN, you’d like to write something like (And I’m just guessing here) 10 must-know real estate terms. Possible, no? Especially if the page linked to at the time doesn’t mention anything about real estate terms but does offer content like 7 Ways to save for your… and 8 Tips for and so on.

Well, making such a link, talking about an actual service you don’t provide, but pointing to other content in the same general area, may not be something you or I would do. That would be a misleading advertisement of sort, perhaps. Decent companies don’t do that.

What decent companies, like Microsoft, would do, is create a link to

10 most-know real estate terms

Stands to reason, no? I mean, you’ve got various 10 most-wanted lists, so it makes sense to have a 10 most know list. Nothing strange there at all. Nobody who gets paid to write these things can be expected to notice that something is slightly off there, now can they?



And maybe it also means that nobody on their payroll bothers to read those links when if they log in to the Hotmail accounts? I saw the same link with the same text there for hours. They should perhaps take this like a focus group result, or usability review, and cancel the whole thing… Saves everyone time and money…

So Download.com are running a poll about switching to FireFox.
And people seem surprised about the high percentage of people claiming to have already switched.



I use FireFox myself. That’s not relevant. What’s relevant is that the poll is… well… not relevant. For pretty much the same reasons most online polls aren’t, but not just.



The result are already skewed by self selection of participants. Even assuming the crowd that goes to the site, download.com, is representative of the global Internet users population, not everyone will answer the poll.

And those that answer are not randomly distributed. Most users won’t answer the poll unless they give a damn about what it’s about.

Many FireFox users today probably do, or they wouldn’t be using it. Most IE users, on the other hand, may not really have a clue what FireFox is, or what it means to them, and so have no opinion. This means that even if the poll had a FireFox?! What’s that? option, which it doesn’t, they wouldn’t bother answering.

An interesting extra data item could be the number of page hits that didn’t access the poll. But that doesn’t mean much by itself as well, since it includes people that didn’t scroll down enough, and the like.

Not to mention repeat hits, automatic scripts, and all the nice other things that usually tend to come up when someone actually pretends to take a poll seriously



In addition to that, you have various blogs and sources sending people to take the poll, such as Spread FireFox. And in fact the large majority of them (I’d say all, but I don’t have the time for a really thorough check, and one may always crop up) are very FireFox friendly. I wager you won’t catch too many Stick with IE sites telling people they can go and vote to switch to FireFox only When heck freezes over.



Besides, they can check (well, try to. Changing the user agent string and the likes is far from impossible) what browser the users used. If someone already uses FireFox, they don’t really need to vote…

Well, if someone uses IE5 they probably don’t need to vote as well ;-)



And then of course, even if the result was both reliable and representative (instead of 0 out of 2), this specific poll is still pretty pointless. It doesn’t help anyone. It doesn’t have a clear purpose.

It’s not asking what browser people are using, apart from one answer saying FireFox, and several others saying Not FireFox.

It’s not asking what features are wanted before switching to FireFox, apart from just one answer mentioning security testing (and yes, it is an issue. Of course more problems are found more frequently on IE, but Mozilla/FireFox do have their share. Am I more worried about watching a JPEG on my browser, or about watching a BMP on it? Though call…). Other reasons aren’t mentioned, so it doesn’t help any of the developers, any of the competitors, or anyone else trying to find trends..

I just can’t seem to find a good reason for the poll at all. The survey result teaches us nothing. No, wait, scrap that, it probably teaches us that CNet are about to cut some staff, since apparently they have some poll writers on the payroll with nothing to do on their hand. Interesting idea, running a poll instead of running a press release and being done with it…



And yes, I voted. What better way to see if the silly results are really like all the comments claimed?