Posts tagged ‘bug’


Not alone in discovering Bing is broken

24.07.2022


MIA again on Bing: Lucire’s home page. The alt tags are not missing, with perhaps some exceptions for small logos. And not having an H1 tag is not fatal to other pages of ours that have been indexed. It remains bizarre.
 
After Holly Jahangiri’s very useful feedback to my previous post, I thought I’d give the search engines she sampled a go for site:lucire.com.

Bear in mind that Duck Duck Go, Ecosia, Qwant, Onesearch and Swisscows license from Bing, and Startpage picks up Google, so their indices will reflect the mothership.

Here’s how we look today. Bing remains well and truly beaten by Google, Mojeek, Baidu and Yandex.
 
Google: 6,100
Mojeek: 3,569
Swisscows: 498
Baidu: 201
Startpage: 198
Virtual Mirage: 100
Yandex: 94
Bing: 50
Qwant: 50
Duck Duck Go: 49
Ecosia: 49
Brave: 14
Searchencrypt: 8
Searx: 0
Onesearch: blocked in New Zealand
 

I am not alone, it seems. This thread on Microsoft Answers was enlightening. Others in the thread have found themselves gone from Bing (but not Google), and Microsoft appears to know about it, admitting to some fault and escalating the issues internally, but nothing ever gets done.

I had that old theory, blogged about previously, that computer databases get worn after a while. I saw that with Vox, a lot of Facebook’s ills can be put down to it, and maybe Bing has now got there? No tech ever wants to admit it because of how crazy it sounds. But if we can lose data on hard drives and USB sticks, then I don’t care how many back-ups these big firms have, they are still fallible. (What if faults in one database are copied on to another, and the checksums weren’t verified?)

I replied to the Microsoft poster, and it’s a pretty good summary so far:

Hi EbinVThomas, here’s my experience, and I’ve run websites for three decades. The short version is I think Bing is stuffed and it’s not a Microsoft core business, so it doesn’t get much love (indeed, one of their FAQ pages has a heading about ‘seach’). I know the Microsoft fans will attack me for saying this, just as the Apple fans have a go at me when I say something negative about Macs, but I haven’t read anything to change my opinion.

We started vanishing from Bing earlier this year, maybe about three months ago. For some of our sites, I thought it was our belated switch to HTTPS for some of them, but as you’ll read, that wasn’t the case.

These sites date from (at their present domains) 1995, 1997, 2002 and 2008, and they are well linked, well respected, and one has been winning awards from 1997 to today. Google and Mojeek have no problems with any of them. Two of the sites (the 2002 and 1995 ones) did drop from their number-one and two positions on Google (for a name search) when they switched to HTTPS but one has mostly recovered, the other (from 1995, with a lot of inbound links to HTTP) fluctuates.

One of the other sites uses Duck Duck Go for its internal site search (and has done since the 2010s), which is powered by Bing. Earlier this year—say about six weeks ago—I noticed that the internal search was getting more and more useless, even though I knew the articles used to be found by DDG.

I began doing site:domain.com searches for this one. It had c. 50 entries on Bing, down from several thousand earlier in the year.

My first reaction was to blame ourselves—maybe it was the full switch to a secure server (some earlier pages were already on HTTPS), or something else. We also began using Cloudflare again after a 12-year break around this time.

I signed up to Bing Webmaster Tools. The site promptly went down to 10 entries! In other words, signing up to Tools made the site’s presence a lot, lot worse.

I found some weird site maps that I never put in, nor did any of my team. Nevertheless, I put in new, fresh ones last week, all pointing to HTTPS. Most of the pages have not been indexed.

I had to turn off Cloudflare’s IndexNow because it was sending some totally irrelevant and old pages and files to Bing. (So we can blame Cloudflare for some issues, but the majority still rests with Bing.)

Since the new site maps, Bing is now returning 53–5 entries (depending on the hour).

It finally included the home page which had been missing from the site: searches. Yet only yesterday Webmaster Tools said the page was not indexed because of certain issues, but it had been found in 2018. That made no sense as it was present until quite recently. Those issues included a description tag being too long (fine, I edited it), and no H1s (but why should there be? Not everyone wants humungous type on their page). But Bing had been fine historically with the page (since Bing started, so well before 2018) and it even appeared in the index during the last few weeks. A related page for our business doesn’t have H1s, an even longer meta description, and it’s on Bing. (It’s just not been entered into Webmaster Tools, which seems to be a kiss of death!)

Webmaster Tools even said it had accepted the site maps and the thousands of pages listed.

As far as I can make out, Webmaster Tools says one thing but reality says another.

So, was it Cloudflare and HTTPS that had knocked us? Well, no. Of the four sites I mentioned, we didn’t change the set-up of the one started in 2008. It’s a reference site, and has plenty of inbound links from Wikipedia since it’s fairly authoritative.

No Cloudflare, and still on HTTP. All fine on Google and Mojeek.

Also thousands of pages.

On Bing: 51 pages.

Thousands of entries have vanished since earlier this year, and I’m going to hazard a guess to say it began happening around the time you wrote your original post.

It has had a slight impact on our traffic, especially since we had promoted Duck Duck Go so heavily since 2010 and encouraged others to shift from Google to it.

It seems that Bing can now only cope with 50-odd pages from certain sites. The older sites have fewer pages indexed now on Bing than they did on Excite or Hotbot in the 1990s, and certainly far fewer than Altavista! Our sites are so incredibly varied—static, dynamic, HTML, PHP—so it can’t be structural or the way we have set things up. None have had issues at Google other than one that dropped in the index for a certain relevant search, and Mojeek is fine with them all and took the HTTPS shift for three of them in its stride.

These are such old sites with a history in Bing, so my feeling is that a new site won’t stand much of a chance.

This is a long way of confirming your original post: it’s not you, it’s them.

Tags: , , , , , ,
Posted in internet, technology, USA | 3 Comments »


Facebook admits we’re experiencing a bug preventing us from managing Lucire’s page

08.05.2022

Last night’s hour-waster was chatting to Facebook Business Support. No, that’s unfair. I was actually assigned an incredibly good rep who took me seriously, and concluded that Facebook did indeed have a bug which means, of all the pages I can manage, the one for Lucire is alone in not allowing me, or any of its admins, to do anything. How coincidental, after losing Instagram and Twitter for periods during 2021.

Ironically, one editor can—of course someone who is supposed to have fewer privileges can do more. Such is Facebook.

A few things I learned. There’s a Meta Business Suite, which a whole bunch of pages got shoved into, whether you wanted it or not. My public page is there, for instance. It seems if you have Facebook and Instagram accounts for the same thing, you’re going to be in there.

Despite the two-factor authentication discussed in the previous post, I actually can get into the Business Suite, via another page I administer for a friend. From there I can get to Lucire’s tools.
 

 

I don’t need two-factor authentication for any of the other pages in there, including my own, and have full access.

Trisha, or Trish as she said I could call her, walked me through the steps, and asked me to get to the Suite page. Then she asked me to click ‘Create ad’, and I get this:
 

 
She asked me to check the account quality, and of course there are no issues:
 


 

She wrote: ‘Thanks for letting me know. It’s weird because I have checked all your assets here and it looks good. But, here’s what I suggest, Jack. We’ll need to report this to our Internal team so they can investigate. You might experience a bug or glitch.’

I theorized: ‘Just so you know, this page dates back to 2007 so maybe it is so old that Facebook’s servers can’t handle it?’

It wasn’t something she responded to, as she stayed on-subject, but it’s a theory worth entertaining, as it wouldn’t be the first time I’ve witnessed this.

So, for now, the one team member who can still go on Facebook for us posted this at my request:
 
All Lucire admins, all automated gadgets sending links to this page, and all Facebook-approved reposting sites, were blocked by Facebook on April 25. Therefore, till Facebook fixes this, there will be no more regular updates to this page other than a limited amount from one of our editors.
 

I doubt they’ll ever fix it, and two years ago I did say I wouldn’t really bother if Facebook went buggy and prevented us from updating again. Clearly I am bothering, as I know we have readers who use Facebook. But I have very little faith this will ever be fixed, since I have seen other reported bugs (some covered on this blog) get ignored for years, and this isn’t a fleeting bug, from what I can make out.

The lesson, as I have probably hinted at more than once, is never rely on a Big Tech service. The sites are so unwieldy that they get to a point where no one knows how to fix them. If earlier experiences are any indication, such as what I experienced at Vox, we have arrived at the end of Facebook pages.

Tags: , , , , , , , ,
Posted in internet, publishing, technology | No Comments »


Facebook pages no longer immediately update after you post

21.12.2021

With Lucire’s Twitter restored—it was a huge distraction over the last two months with various automatic posting gadgets needing to be reprogrammed, and the Twitter-to-Mastodon cross-poster still does not work (it’s what happens when you modify things that are working perfectly well: it’s impossible to put them back)—we wanted to get some of our social media back up to speed.
   So let’s get back to rubbishing Facebook, shall we?
   Because whenever we post, whether it’s through another program or directly on Facebook, the post just does not show on the page itself, unless you’re on the ultra-slow mobile edition (m.facebook.com) where you’re likely to give up before the posts begin to load. You have to wait many hours, even a day, for something to appear on the full-fat web version.
   This reminds me of those days when our Facebook walls stopped updating on the 1st of each month, presumably because someone in Menlo Park had to flip a switch to tell the website that the new month had begun. And they wouldn’t do it till it hit midnight in California since everyone on the planet must live in the one time zone.
   What’s the bet it’s a related glitch that existed at the end of 2011 and the beginning of 2012, but we need to wait till midnight in California for Facebook to know it’s a new day?
   Here’s the mobile version (albeit viewed on the desktop) earlier on Tuesday:

And here’s the web version, still unchanged 12 hours later, with the newest post after the pinned one from December 18.

   Lesson: don’t use Facebook if you wish to tell your audience something urgently. You are better off sending an email: they’re more likely to see it in a timely fashion. And if your following isn’t that big, and you need your fans to know something urgently, it might actually be quicker to telephone them! Social media—forget it.

Tags: , , , , , ,
Posted in internet, marketing, technology | No Comments »


InShot Music Player is also forgetful, though it’s not as bad as the last two

20.08.2021

One bug that creeps up at unpredictable intervals with InShot’s Music Player—though it is not as severe as the bug on Muzio Player—is that after a while, it forgets that it should shuffle the tracks and resorts to alphabetical order, starting from the top.
   Considering this isn’t something that has affected any other music player, I find this very surprising.
   These four screenshots were taken between July and August of the recent tracks. There’s no rhyme or reason the player would suddenly go to the top of the list, but when I begin hearing the same sequence of tracks, I know something’s not right. And it has been happening since I installed the player, though the first couple of times I didn’t realize it was a bug.
   I would tell Inshot directly but my last (highly positive) email went unanswered, so a public blog post is the next best thing, in case others have come across this bug.
   With how forgetful computer programs are all the time, including the player I had on my phone prior to this, I wonder: should I invent the ini or preference file? It seems that in this universe, these haven’t been invented yet!




   On a side note, Meizu’s native music player has also forgotten to show the list of tracks, which remain linked after my herculean effort earlier this year. Its search still fails to scan the SD card.

Tags: , , , , , , ,
Posted in design, technology | No Comments »


Fixing Wordpress’s problem of fake bolds and italics

11.06.2021

I haven’t been able to find anything on this bug online, but it’s very common.
   As far as I can recall, all of our online publications that use Wordpress have themes designed or modified by yours truly. However, Lucire Rouge has a mostly bought-in theme, where my changes have been limited to a couple of CSS rules. The theme developer actually came in and helped us with a few modifications, which shows the extent to which he does follow-up for paying customers.
   But there was one thing he was never able to crack, and I don’t think it’s his fault, since it happens on a lot of websites, including Medinge Group’s (also a theme I did not design, though I did earlier ones). On both these sites, there were no bolds and italics. There still aren’t on Medinge’s.
   There are <strong> and <em> codes in there, but the bolding and obliquing are done by the browser. The font files actually aren’t loaded, so what we see are false bolds (the browser attempts to “overprint” the roman, duplicating the outline and shifting it marginally to give the illusion of a heavier typeface) and obliques, not italics (it’s the roman file pushed over 15 degrees or so). The former is particularly bad, as the outlines clash, and the result can be hollow glyphs, something that any font developer will know when one outline winds up accidentally on top of another in Fontographer or Fontlab.
   These Wordpress themes rely on Google Fonts (another sin, in my opinion) so I don’t know if the fault lies with Google or Wordpress, or the developer. If Wordpress does indeed power 70 per cent of websites, then I have to say the bug is awfully common, and I probably do see it on a very high percentage of visited sites.
   The themes allow us to select the font family, but the selection only calls a single font file from the family.


Above: A graphic clipping text from Lucire Rouge that I sent to the developer.

   The solution, as I discovered after months of toing and froing with Lucire Rouge’s theme dev, was to do your own font-linking rules in the CSS file and upload the fonts themselves to the relevant directory on the server. I must note publicly the ‘months’ were not his fault, but due to my own delay. I should not expect computer programmers to be typographers, either.
   It is something that one needs to watch out for, as the fake bolds and italics are horrible to look at, and must look amateur, even to the non-professional.



Above: Fixed at last by yours truly.

Tags: , , , , , , , , , , , , ,
Posted in design, internet, media, publishing, technology | No Comments »


On OneDrive, Flickr, and FLOC

19.05.2021

Yesterday, I worked remotely, and I don’t know what possessed me, but as OneDrive was activated on my laptop, I decided to save a word processing file there, planning to grab it from my desktop machine later in the day.
   Normally I would just leave the file where it was and transfer it across the network, which is what I should have stuck with.
   Heck, even transferring a file using a USB stick would have been a better idea than OneDrive.
   I hadn’t signed up to it on my desktop PC. I went through the motions, used the default settings where it said it would back up documents and pictures (while making it clear my files would remain exactly where they were). I grabbed the file I need—the entire 18 kilobytes of it—and thought nothing more. I deactivated OneDrive as I saw no real use for it any more.
   Bad idea, because most of my desktop icons vanished, and my Windows default documents’ and pictures’ folders were emptied out.
   After reactivating OneDrive, I found the lot in the OneDrive folder, and promptly moved them back to their original folders. The desktop files—the text files I had on there plus the icons—I duplicated elsewhere. Ultimately, I made new shortcuts for everything—thank goodness my laptop’s icon layout is identical to my desktop’s—and restored the three text files from their duplicate directory.
   The above took me all of a few minutes to write but in reality I spent an hour fixing this—something that Windows said would not happen.
   Chalk it up to experience—consider this fair warning to anyone who thinks of using “the cloud”.
 
 

Also in the “say one thing, do another” file for yesterday: I attempted to sign in to my Flickr account, which has not been touched since around 2008. I tried a range of addresses I had in 2006, when I originally signed up, and attempted to do password resets. Flickr: ‘Invalid email or password.’ I even tried an address that Yahoo! emailed me at in 2018 concerning Flickr, and which Flickr itself said might be the correct email (use your Yahoo! username and add ‘@yahoo.com’ to the end of it).
   I had no other option but to email their support, and mentioned that I was a paying Smugmug customer, given that the photo site now owns Flickr.
   They have responded in a timely fashion, not telling me the email I had used, but said they had sent it a password reset in there.
   Surprisingly (or maybe not, considering we are talking about another big US site again), the address was indeed one of the ones I had tried (I’m glad I kept a record). Except now it works—what’s the bet that post-enquiry, they fixed things up in order to send me that reset email?
   I thanked the support person for the reset email, but suggested that they had some bugs, and fixing them would mean less for him to do.

Don Marti linked an interesting article in The Drum in which he was quoted. Duck Duck Go, Firefox and Github have all opposed Google’s new FLOC tracking method. Meanwhile, Bob Hoffman points out that only four per cent of Apple users have opted in to tracking after the Cupertino company’s new OS opted you out by default.
   Most of the time, people tell me that they find targeted ads ‘creepy’ as they appear from site to site, so it’s no wonder that take-up has been so low with Apple users. So if not FLOC, then what?
   Well, here’s a radical idea: show ads on sites that have subject-matter relevant to the advertiser. It’s what happened before Google’s monopoly, and there were plenty of smaller ad networks that did a great job of it. The prices were still reasonable, and Google wasn’t taking a big cut of the money earned. Of course Big Tech doesn’t like it, because they won’t earn as much, and the old system actually required people with brains to figure out how best to target, something creepy tracking has tried to replace.
   The old methods, with their personal touch, resulted in some creative advertising work—I remember we had some page takeovers on Lucire’s website where the traditional header was redesigned to show off the R55 Mini, thanks to one of our earlier ad directors, Nikola McCarthy. No tracking involved, but a great brand-builder and a fantastic way for Mini to get a fashion connection. Ads with tracking are so transactional and impersonal: ‘Buy this,’ or, ‘You’ve searched for this. Buy this.’
   I doubt it does the brands much good, and before you say that that doesn’t matter, let me also add that it can’t do the humans much good, either. The user’s purpose is reduced to clicking through and buying; so much for building a relationship with them and understanding their values. That isn’t marketing: it’s straight selling. Which means the marketing departments that put these deals together are doing themselves out of a job. They’re also spending money with a monopoly that, as far as I have read, doesn’t have independently certified metrics, which 20 years ago would have been a concern with some agencies.
   I do like innovations, but every now and then, I feel the newer methods haven’t done us much good. Tracking is tracking, no matter what sort of jargon you use to disguise it.

Tags: , , , , , , , , , , , , , , , , , , , , , ,
Posted in business, internet, marketing, media, New Zealand, publishing, technology, USA | No Comments »


NewTumbl takes things seriously

05.01.2021

I have to say I’m impressed with NewTumbl as they responded to my Tweets about potential censorship and post moderation.
   I think they will allow me to share a few points.
   First, they took me seriously. The fact they even bothered to look into it is well beyond what Yahoo, Amazon, Facebook and Google would normally do, and I’m talking about Yahoo in 1999. They also answered every point I made, rather than gloss over or ignore some. Out of their thousands or myriads of users, they were actually good enough to deal with me one on one.
   Secondly, they assure me there’s no censorship of the kind I suspected but think a temporary bug could have been behind Mbii’s inability to see my posts. They will delete illegal content, and that is the extent of it.
   Thirdly, if I may be so bold as to say this one, my understanding of the posts’ levels is correct and those moderators who objected to my content are incorrect.
   I won’t reveal more than that as some of the content refers to future actions.
   I’ve said I could put a toe in the water again over at NewTumbl, and, ‘I really appreciate you taking this seriously and certainly it all helps encourage me to return.’
   Being honest and up front really helps.
   The one thing preventing me from heading back in a flash is I’ve become rather used to adding to the gallery here. We’ll see: I felt it was ‘No way, José’ a month ago (although I always maintained a “never say never” position—I mean, it’s not Google Plus) and now it’s more ‘The jury is out.’ At the least I might pop by more regularly to see what’s in my feed.

Tags: , , , , , , ,
Posted in business, internet, technology, USA | 5 Comments »


Looks like Twitter makes up your settings, too

02.12.2020

Speaking of Twitter doing weird things, I checked out some of Lucire’s settings on there today, something I haven’t looked at for a long time.


   I do not ever recall telling them I was in Malaysia—it’s not a country we’ve even had a correspondent in—and Estonian and Welsh were never marked as languages. I’m not even that sure about Romanian since our edition there shut in the 2000s and the Twitter account dates from 2009.
   Given yesterday’s post, I should be able to be more certain: I didn’t put in Malaysia, and I didn’t put in Romanian.
   I will give them one compliment: the advertising preferences were a damned sight more accurate than anything I saw on Facebook (back when Facebook let me see them). I still deleted them though.

Tags: , , , , , ,
Posted in internet, technology, USA | No Comments »


Confirmed: Twitter does unfollow people

01.12.2020

Twitter bird fallen
Pixabay

People have long suspected that Twitter “unfollows” without human intervention (I’ve heard the stories for over a decade), but I never had any first-hand proof. Like the Google Ads Preferences Manager (they don’t use apostrophes in Mountain View) situation in 2011, where I initially doubted myself, thinking maybe I hadn’t opted out when in fact I had, Twitter had me thinking the same thing for years. Maybe I was never friends with that person. The technology wouldn’t have unfriended them—there would be little point. Maybe I only ever saw them in re-Tweets.
   But finally along comes a case where I know for a fact I never unfollowed someone, and he says he never unfollowed me. (Please bear with me while I use the colloquialisms without putting them into quotation marks.)
   I have a Twitter plug-in that tracks unfollowers and was very surprised when a real-world friend, Jordan, showed up on the list. I looked at his account, and, to my surprise, I wasn’t following him. I knew for a fact I was—this was one case where I was sure, and I had even replied to something of his last week. It is, as my fellow RNZ Panellist Andrew Frame put it, ‘Eureka!’ Or, to bring it out of ancient Greece and into language my contemporaries can understand: ‘You’re nicked.’ You can read the thread here.
   So just like Google in 2011—which had likely been lying to the public for two years about opting out through its Ads Preferences Manager—I am right and the technology is wrong. Back then I really had opted out and Google deceptively opted me back in, something which I point out ad nauseam on this blog. Here, neither Jordan nor I unfriended the other.
   Last year I wondered why I wasn’t following Peter Lambrechtsen, another IRL friend, but put it down to my own omission; I don’t believe that to be the case any more.
   You may wonder how hard it is for a site to keep track of who follows whom, but I have long accused Facebook of having databasing problems, based on my experience as a regular user from 2007 to 2017; surely the same thing must afflict Twitter as it scaled to milliards of users. When in doubt, don’t: experience says that when it comes to these big sites, usually it’s not your fault, but the technology’s. Twitter is as fallible as Google and Facebook.

Tags: , , , ,
Posted in internet, technology, USA | 1 Comment »


Search engines favour novelty over accuracy and merit

01.10.2020

I was chatting to another Tweeter recently about the Ford I-Max, and decided I’d have a hunt for its brochure online. After all, this car was in production from 2007 to 2009, the World Wide Web was around, so surely it wouldn’t be hard to find something on it?
   I found one image, at a very low resolution. The web’s not a repository of everything: stuff gets removed, sites go down, search engines are not comprehensive—in fact, search engines favour the new over the old, so older posts that are still current—such as this post about the late George Kennedy—can’t even be found. This has been happening for over a decade, so it shouldn’t surprise us—but we should be concerned that we cannot get information based on merit or specificity, but on novelty. Not everything new is right, and if we’re only being exposed to what’s “in”, then we’re no better at our knowledge than our forebears. The World Wide Web, at least the way it’s indexed, is not a giant encyclopædia which brings up the best at your fingertips, but often a reflection of our bubble or what the prevailing orthodoxy is. More’s the pity.

I can’t let this post go without one gripe about Facebook. Good news: as far as I can tell, they fixed the bug about tagging another page on your own page, so you don’t have to start a new line in order to tag another party. Bad news, or maybe it’s to do with the way we’ve set up our own pages: the minute you do, the nice preview image that Facebook extracted vanishes in favour of something smaller. I’ll check out our code, but back when I was debugging Facebook pages, it was pretty good at finding the dominant image on a web page. Lesson: don’t tag anyone. It ruins the æsthetic on your page, and it increases everyone’s time on the site, and that can never be healthy. Time to fight the programming of Professor Fogg and his children (with apologies to Roger McNamee).



Top: The post Facebook picks up from an IFTTT script. Above: What happens to a post that once had a proper image preview after editing, and tags added.

Tags: , , , , , , , , ,
Posted in cars, culture, interests, internet, technology | No Comments »