In a few cases, the start dates are well-informed guesses
May 4, 2013 9:25 AM   Subscribe

Predicting Google Shutdowns. "In the following essay, I collect data on 350 Google products and look for predictive variables. I find some while modeling shutdown patterns, and make some predictions about future shutdowns. Hopefully the results are interesting, useful, or both." Gwern exhaustively analyzes Google products past and present with an eye to establishing what's not long for the bitverse. tl;dr? Results.
posted by mwhybark (88 comments total) 16 users marked this as a favorite
 
Naturally, we are preached at by apologists that Google owes us nothing and if it’s a problem then it’s all our fault and we should’ve prophesied the future better

If people seriously do this then they should go to extra hell. Offering a product, even for free, implies that it's going to have some degree of continuity. Neither time nor brain space is valueless, and asking them to learn to use something only to take it away is willfully wasting both. Pulling it out from under people will rightfully piss them right the hell off, and the more successful the product the more you'll make them mad. Google lost a ton of goodwill over the Reader shutdown fiasco, I know that at least I will never give them so much credit again.
posted by JHarris at 9:32 AM on May 4, 2013 [9 favorites]


I wonder how well his model would have predicted the shutdown of Reader?
posted by Foci for Analysis at 9:36 AM on May 4, 2013 [2 favorites]


Offering a product, even for free, implies that it's going to have some degree of continuity

According to what rule, ever?
posted by yoink at 9:47 AM on May 4, 2013 [8 favorites]


The Google Glass survival curve is interesting, with a 31 percent chance of making it past five years. Seems surprising that this patient is given such poor odds, given the amount of pre-release marketing Google is doing with it, and how it seems like such a focused pet project for one of the company's founders. I'd be curious to know the answer to Foci's question, too.
posted by Blazecock Pileon at 9:51 AM on May 4, 2013


If people seriously do this then they should go to extra hell. Offering a product, even for free, implies that it's going to have some degree of continuity.

Bullshit. Every product team, in every organization, has to balance the cost of the offering vs the return they'll get for it, and consider if the resources spent on the product could be better used elsewhere. Sometimes that sucks for the users. That's life.
posted by NotMyselfRightNow at 9:52 AM on May 4, 2013


Sometimes that sucks for the users. That's life.

It also sucks for historians, too, which indirectly sucks for everyone, users and non-users alike. The author mentions the loss of massive amounts of Arab Spring data, for example, which was just evaporated without consideration for future utility.
posted by Blazecock Pileon at 9:55 AM on May 4, 2013 [4 favorites]


Does not look good for google voice, which would righteously screw current users. You think it sucks when Reader shuts down, just wait till Google breaks your phone.
posted by leotrotsky at 9:55 AM on May 4, 2013 [4 favorites]


So, whether you're bitter or sanguine about the Reader shutdown, this "analysis" is mostly mathematical crazy talk. The presence of numbers and graphs doesn't mean it's actually meaningful. Just as the Apple Kremlinologists attempt to read the tea leaves to predict product releases, this is basically the same thing for product EOLs.

Also, every product gets EOL'ed. It's only a question of when.
posted by GuyZero at 9:58 AM on May 4, 2013 [3 favorites]


The lazy half of me is glad not to see Google Code on the model's chopping block, given the work it takes to distribute our stuff to end users. The other half is not so glad, as it would be good motivation to get our stuff off Google Code and on to a more reliable service. Overall, I think it will be the lack of warning of the end coming that will screw us the most.
posted by Blazecock Pileon at 9:59 AM on May 4, 2013


On one hand, I'm glad this isn't forecasting doom for Google Apps, as other predictors have. On the other, predictions of doom for Google Voice make me pretty unhappy, as I use it as a free alternative to paying for "visual voicemail" services.
posted by limeonaire at 10:01 AM on May 4, 2013


I'm already sunsetting my reliance on google voice. The last thing for me to do is to port my phone number over to Voipo.
posted by wotsac at 10:01 AM on May 4, 2013 [1 favorite]


According to what rule, ever?

According to the rule of DON'T DO IT OR WE HATE YOU. We're allowed to hate here; as the aggrieved, it's our duty to hate. (We also hate those who try to somehow defend this behavior, BTW, as enablers.) Offering a service implies it'll be around for some time; suddenly taking it away debits the respect we gave your company for offering it in the first place (in the case of Reader this was considerable), plus an additional fee for wasting our fucking time.

Hypothetical example: what if Google decided that Gmail wasn't profitable enough and killed it? You'd be right there saying HAW HAW you fools relying on a free service. But what if you can't afford to pay for email? Your attitude is basically rich man's chauvinism; lots of people don't have the cash to outlay for every damn basic internet service.
posted by JHarris at 10:03 AM on May 4, 2013 [12 favorites]


> Google lost a ton of goodwill over the Reader shutdown fiasco, I know that at least I will never give them so much credit again.

Google is now a publically held corporation and under the same pressure to cut cost and increase profit and make numerical forecasts for the next quarter as a dog food company is. They care about our goodwill. Goldman Sachs is ahead of us in line and the don't be evil boys will serve Kapital or be out before you know it.

(Gwern's article is very good. I read it a few weeks ago in the Reader cancellation brouhaha. The numbers most salient to me:

Relative risk versus average:

search .05
adsense .06
gmail .08
glass .10
chrome .24
blogger .32

My own opinion is glass is doomed and chrome's viability relative to blogger is grossly underestimated. But this is the kind of an article that cannot be digested in one sitting. I will be returning to re-read it many times I predict!)
posted by bukvich at 10:07 AM on May 4, 2013 [2 favorites]


Never long in these threads before folks show up to tell us that not only should companies be able to do whatever they damn well please, they should also be able to do it completely free from any criticism at all, huh
posted by ominous_paws at 10:07 AM on May 4, 2013 [17 favorites]


Well, I mean, they are able to do whatever they please, basically. I don't think anyone is saying they should be immune from criticism.
posted by whir at 10:10 AM on May 4, 2013 [2 favorites]


Google Voice doesn't make a profit, so its chances of succeeding are low...except that they have hinted that they will start charging next year (by stating that it will remain free in 2013).

If they start charging, then it becomes profitable, then the prediction changes. (Assuming everyone doesn't abandon it).
posted by eye of newt at 10:10 AM on May 4, 2013


Google is now a publically held corporation and under the same pressure to cut cost and increase profit and make numerical forecasts for the next quarter as a dog food company is. They care about our goodwill.

They should. It was that goodwill that built the company to begin with, a large pile gained from offering extremely good, principled search in the web's infancy, while other search engines were racing each other to find new ways to monetize paid search placements.

But it's all worth anything at all because we, as in the internet as a whole, respect them for it. This is actually quite fragile. There is actually not a whole lot separating Google, the internet behemoth everyone adores, and Yahoo, the undead malingerer, punchline of the web.

Well, I mean, they are able to do whatever they please, basically.

They wouldn't be able to if there were meaningful regulatory oversight.

I don't think anyone is saying they should be immune from criticism.

HA.
posted by JHarris at 10:18 AM on May 4, 2013 [4 favorites]


JHarris: you can see it here and on Hacker News without fail. It's the geek version of blaming the victim.

Foci: It's a bit late for predicting on Reader, yeah, but as it happens I did run the 5-year stuff on all the entries (it was easier than just for the live entries) and one can pull them out of the CSV; the estimate (5 years from June) was 67%. I don't know if this is reasonable or not because of hindsight.

Blazecock: I agree that the Glass estimate seems questionable. But how do you quantify our intuition in terms of actual data? There's no 'importance' variable floating around we can just copy down and add to the regression.

leotrotsky: Let's hope Google keeps that in mind! On the other hand, wasn't there at least one business which had to shut down after Reader did?

GuyZero: It'd be nice if you could be more specific in your criticisms.

eye of newt: Just so.
posted by gwern at 10:21 AM on May 4, 2013 [3 favorites]


They wouldn't be able to if there were meaningful regulatory oversight.

I'm curious about what sort of regulations you think should be placed on Google that would have prevented them from shutting down Reader. I mean, I miss it too, but it's not like Con Edison suddenly stopped delivering gas to my house or something.
posted by whir at 10:25 AM on May 4, 2013 [2 favorites]


except that they have hinted that they will start charging next year (by stating that it will remain free in 2013).

Yeah, the thing about hinting is it means shit. MAKE A DAMN EXPLICIT STATEMENT WE CAN RELY UPON. This pussyfooting around is a way of hedging bets, of saying nothing under the guide of saying something. It's rhetorically null: if they decide not to charge next year, then they'll say it'll remain free in 2014; if they decide to charge, they'll make one of those statements like

We're proud to announce Google Voice With Lemon!
Due to the high cost of citrus Google Voice will now cost $3/mo.

posted by JHarris at 10:25 AM on May 4, 2013 [1 favorite]


What I don't get is why they didn't sell Reader to someone they thought could make a profit on it, or at least give some warning and offer some sort of export, like they did with Google Wave. Or you know, add more ads and make it profitable.

whir: A fair-warning law and something allowing for data export (30 days warning, and a guarantee of data portability perhaps?)
posted by Canageek at 10:26 AM on May 4, 2013


Canageek: an insider said that Reader was too caught up in the Google infrastructure to feasibly split out (quoted in my page); I would believe this since you can tell just from the papers they release that they have a whole array of interesting services and ways to architect things. And Google has long offered the ability to export your subscriptions as an OPML, and IIRC it exports some other stuff like starred items. What people are bitching about is stuff like the loss of the Reader API or the unique archive of RSS feed items stretching back to ~2005 including many deleted or dead sites (which you cannot export).
posted by gwern at 10:29 AM on May 4, 2013 [2 favorites]


Google Voice doesn't make a profit, so its chances of succeeding are low...except that they have hinted that they will start charging next year (by stating that it will remain free in 2013).

Don't they say this every year?

At this point, if a "Google Voice requires payment" dialog pops up on Jan 2014, the fact that they "have been warning us" for the the past year means nothing. It will still be unexpected for many, and vocal parts of the Internet will revolt.

At least with Google Voice, you would have the option to pay. No such luck for Google Reader fans.
posted by meowzilla at 10:31 AM on May 4, 2013 [1 favorite]


I'm curious about what sort of regulations you think should be placed on Google that would have prevented them from shutting down Reader.

How about something about offering a public product and getting everyone to fucking rely upon it, then shutting it down with three months notice. If you offer a public service in the first place you should at least offer some decent time before killing the damn thing. And the rule shouldn't be swayed by calling the service beta either, because half the web is in perpetual beta.

I'm no lawmaker, but if we're talking practicality then sure I'll have a crack at it. At the very least there should be offered an optional guarantee of continuity; a promise, with legal weight behind it, that an offered service won't be shut down for a year. That way people will at least have some way to distinguish between those services that will be around for a bit and thus can risk building them into their workflows, or those that they might pull the rug out from under on a moment's notice.
posted by JHarris at 10:33 AM on May 4, 2013 [1 favorite]


A fair-warning law and something allowing for data export (30 days warning, and a guarantee of data portability perhaps?)

Ok, well I don't want to derail the thread too badly, nor set myself up as Google's white knight since I'm quite skeptical about a lot of things they do. But I mean, have you had fair warning yet? Google removed the link from their main nav on the home page, but Reader is still there, functioning the same way it always did, and you can use google's export tools to get your data out and into feedly or whatever pretty easily.
posted by whir at 10:34 AM on May 4, 2013


Somebody doesn’t seem to understand Capitalism and Free Markets. Be angry at the right people for the right things.
posted by bongo_x at 10:36 AM on May 4, 2013 [1 favorite]


the list of the 10 least risky (increasingly risky order):

Search


I don't know about that. Search probably isn't the most profitable. I can imagine a future where Google retires Search.

And then we'll have to go back to doing it the old way I choose my keywords and ask a friend about them.
posted by twoleftfeet at 10:40 AM on May 4, 2013


At the very least there should be offered an optional guarantee of continuity; a promise, with legal weight behind it, that an offered service won't be shut down for a year.

What company would be dumb enough to offer such an agreement without something in return?

The problem with Reader is that no one paid for it, including all the application developers who used it as a backend. In days gone by, pricing something for free would be considered some kind of product dumping. Now we're complaining when a company won't offer their product for free, forever? Times have changed.
posted by meowzilla at 10:41 AM on May 4, 2013


It's exactly because free is dumping that cutting it off is so pernicious. They wiped out the whole ecosystem by charging an unfair price for their product, and only then did they discontinue it.
posted by wotsac at 10:46 AM on May 4, 2013 [20 favorites]


But I mean, have you had fair warning yet? Google removed the link from their main nav on the home page, but Reader is still there, functioning the same way it always did,

It's only been a month since the shutdown was announced. It won't be in two more. That's less than a year, according to sainted Ms. Medlin's 1st grade arithmetic. People's habits and workflows don't easily change overnight or even after a month; we try to construct some kind of rhythm for using the internet, so we can work efficiently. In my case, Reader is (soon to be was) a major resource in finding Metafilter post topics; it directly harms my post-building ability. There might not be a replacement for it out there, like ever, there is no law of the internet or of fate that demands there be, but at least with a year's notice there would be a greater chance that someone can meaningfully fill that niche before the shutdown.

and you can use google's export tools to get your data out and into feedly or whatever pretty easily.

At least Google offers that; according to the logic of the corporate behavior defenders, they shouldn't have to do that!

And I'd just like to add that Feedly's interface sucks, and not everyone has the resources to run TinyTinyRSS, nor the technical know-how to set the darn thing up.
posted by JHarris at 10:49 AM on May 4, 2013 [2 favorites]


The entitlement is amazing. I don't want to even think of the shitstorm that will come up when Metafilter shuts down. If people balk at the loss, of a free, non-ad based service, what will they do when the amazing sum of $5 is on the line.

Hell, metafilter doesn't have a plan. They've joked about it anytime the topic has come up. So does metafilter need regulation?
posted by zabuni at 10:49 AM on May 4, 2013 [3 favorites]


Shouldn't we be celebrating the return of competitive products, then? Possibly a company who would advance RSS as an open standard, instead of leaving it as an afterthought while setting up knolls and +Pages?
posted by meowzilla at 10:50 AM on May 4, 2013


You can't really predict this stuff from usage stats or even revenue.

Google has effectively unlimited money, but it is always short of engineers. Teams compete for the new hires, but there aren't many because the bar is so high. Acqui-hires often come with a product attached (e.g. what became Google Voice) and aren't up for grabs, at least at first. Competition for the existing engineers is therefore fearsome.

Even a successful product can get killed if it has a particularly desirable engineering team and a higher-priority project wants those people.
posted by w0mbat at 10:51 AM on May 4, 2013 [4 favorites]


What company would be dumb enough to offer such an agreement without something in return?

What they would get in return is more people would use the service. In fact, if this caught on, people would tend to look askance at any service that didn't offer that guarantee. "Wait, they won't even promise this thing will be around for even a year?!" Y'see?
posted by JHarris at 10:52 AM on May 4, 2013


By saying Google should at least offer people the chance to pay for the free things they've become dependent on rather than simply discontinuing them, we're admitting that if Google acted like your local heroin dealer, that would be an improvement.
posted by jamjam at 10:54 AM on May 4, 2013 [7 favorites]


The most interesting part of this discussion to me back when the Reader shut down was announced was how to profit in the next cycle of google culling. If you had a scalable cheap Reader clone ready to roll out you could make a fortune. It's like when lions take out a huge antelope there are at least twenty vulture meals left on the carcass when they finish.
posted by bukvich at 10:54 AM on May 4, 2013


Touch my google voice and I will burn you with the Hatefire of a thousand Hatesuns, google.
posted by Avenger at 10:54 AM on May 4, 2013 [7 favorites]


UPDATE: Mefi's Own (TM) gwern.

Your OP regrets the oversight.
posted by mwhybark at 10:55 AM on May 4, 2013 [1 favorite]


Shouldn't we be celebrating the return of competitive products, then? Possibly a company who would advance RSS as an open standard, instead of leaving it as an afterthought while setting up knolls and +Pages?

Theoretically, sure, and if Google wants to throw that goodwill away nothing says they have to keep it. Practically though, a lot of people are going to be left in the lurch in a couple of months.

Touch my google voice and I will burn you with the Hatefire of a thousand Hatesuns, google.

YES, YOUR JOURNEY TO THE DARK SIDE IS NEARLY COMPLETE.
posted by JHarris at 10:57 AM on May 4, 2013


Offering a product, even for free, implies that it's going to have some degree of continuity. Neither time nor brain space is valueless, and asking them to learn to use something only to take it away is willfully wasting both. Pulling it out from under people will rightfully piss them right the hell off, and the more successful the product the more you'll make them mad. Google lost a ton of goodwill over the Reader shutdown fiasco, I know that at least I will never give them so much credit again.

The Reader thing continues to be a very confusing move to me, and I'd like to hear at least a reasonable explanation for why Google did this (even in theory). Google has been willing to spend millions and even billions on things that don't provide an immediate financial return, or are simply loss-leaders but garner affection for the company. Was Reader really that much of a resource or money sink that it wasn't worth continuing to cultivate good will, or worth risking ill will by pulling the plug? Was there a benefit/setback analysis of continuing to support Reader that discusses why this was worth it to Google to discontinue?
posted by SpacemanStix at 10:59 AM on May 4, 2013 [1 favorite]


What would also be hilarious is that requiring a year's worth of continuity would basically mean that only companies like Google could offer free services. The like of Newsblur couldn't raise the capital necessary to make sure that they would be able to offer services for a year, in case of bankruptcy or the like.
posted by zabuni at 11:00 AM on May 4, 2013


I uploaded all my data to their cloud, and then they went away.

That's just the way clouds work, they said.
posted by twoleftfeet at 11:02 AM on May 4, 2013 [4 favorites]


Hell, metafilter doesn't have a plan. They've joked about it anytime the topic has come up. So does metafilter need regulation?

Perhaps! mathowie has said he's thought about shutting the site down in the past; it would have been nice to know he couldn't have just flipped the switch and that'd be the end of it. And we pay money to participate here (although in fairness, when he said that it was before the paid signups). My suggestion is only to put a bit of legal weight behind what most people do out of basic fucking courtesy.

But honestly, now we're starting to quibble about what service means. Is Metafilter really the same scale of thing as Google Reader? That's something best left up to the people who actually draft such a law, the people who have experience deciding these kinds of questions.
posted by JHarris at 11:04 AM on May 4, 2013


(sidenote: NewsBlur has become my primary GR replacement, although I actually prefer The Old Reader. NewsBlur's iOS app, while offering plenty of half-baked usability irritations, is somehow more convenient than OR's web UI for me. This user behavior puzzles me. I should note the NewsBlur dev proactively contacted me to request some use-case info after I complained abut some of the app's shortcomings on Twitter and I, prototypical lazy user that I am, have not followed through.)
posted by mwhybark at 11:05 AM on May 4, 2013 [1 favorite]


What would also be hilarious is that requiring a year's worth of continuity would basically mean that only companies like Google could offer free services.

Well again, my suggestion was that companies could offer an optional year of continuity. That way customers could select against those people who for some reason refuse to offer it.

But I didn't mean to derail this thread with my own personal toy idea about how to prevent future Reader-style fiascos. Someone asked; I answered. There might be better ways to do it. The absence of explicit contraindications is not intended to imply that one should not taunt Happy Fun Ball.
posted by JHarris at 11:08 AM on May 4, 2013


Was Reader really that much of a resource or money sink

AllThingsD reported that it was mainly about privacy compliance. Google keeps getting into legal trouble for privacy violations around the world and apparently they need to limit their exposure to regulators as much as possible.
posted by zixyer at 11:13 AM on May 4, 2013 [1 favorite]


Well again, my suggestion was that companies could offer an optional year of continuity. That way customers could select against those people who for some reason refuse to offer it.

A customer is someone who pays for a service. In which case there are usually agreements that require legal notification of service termination, or contracts etc.

Google Reader had no customers. It just had freeloading users.
posted by meowzilla at 11:14 AM on May 4, 2013


I only use Gmail for non-critical things, saving my good old POP3 account for everything else. It's a pain sometimes but it just feels right to own my own data.
posted by tommasz at 11:21 AM on May 4, 2013


If I could pay less than $10 a month for Google Voice, it would still be worth it -- the unlimited texting plan on my phone is an additional $10 that I now don't have to pay because I text through Voice. Having a nifty telephone number is bonus.
posted by fiercecupcake at 11:28 AM on May 4, 2013


A fairy tale in three parts:

Part One:

So, once, rss is shiny and new and lots of little indepenent developers are playing with it.

Along comes Google. By throwing essentially unlimited dev hours, disk, servers, and bandwidth at it they produce a product clearly more robust than what can be produced by any small startup or independent firm.

They offer their robust product for free.

Users, predictably, migrate to use Reader, devastating all competing offerings.

Part Two:

Google continues until it looks like rss usage has reached a high from which it will only fall.

Google pulls the plug on their service, leaving the users to join the independent devs in the trash pile.

Part Three:

We are told that criticism of Google is unwarranted because we just don't understand how free markets work.


And now a moral:

Part of a free market is the freedom to broadcast your assessment of a company's reliability and trustworthiness, and to factor that into your economic decisions.
posted by tyllwin at 11:48 AM on May 4, 2013 [16 favorites]


And always, when the subject of data/service continuity regulation comes up, it gets argued in terms of a small player running out of money and shuttering their service.

And yet, the stories that touch off such discussions all seem to be about companies that are trucking along just fine, they just don't feel like providing a particular service anymore, so they just go drown it in the tub, along with the personal data of those huge numbers of users they were bragging about just a month previous.

It seems to me that these are two different situations that the law could treat differently.
posted by reprise the theme song and roll the credits at 12:03 PM on May 4, 2013 [1 favorite]


The word entitlement is bouncing around my head, related to both sides of the argument.
posted by davejay at 12:07 PM on May 4, 2013


Oh, and I wonder how many companies would offer free services if regulators prevented them from shutting down those services.
posted by davejay at 12:09 PM on May 4, 2013


The word entitlement is bouncing around my head, related to both sides of the argument.

Hey, y'know, it ain't my stuff I'm worried about. My stuff is on my hard drives in my house. I never trusted "the cloud", and what with everything that's happened, I trust it even less these days, and regularly advise other people not to trust it either.

I just don't like seeing other people get fucked.

If that's entitlement, then guilty as charged I guess.
posted by reprise the theme song and roll the credits at 12:14 PM on May 4, 2013 [2 favorites]


I often wonder what would happen if one of these internet giants decides to go all "Galt" one day?

Like, it was one thing when it was free personal email from Hotmail and such. But imagine what would happen to businesses around the world if Gmail, Google apps and Google docs were suddenly inaccessible?

Ok, that one's not *that* bad. What about Amazon S3? The hottest thing in internet now is to host all your assets (images, videos, etc) in "the cloud." What if one day the cloud decides to not be there anymore? You thought banks were too big to fail- wait until one of these companies gets in financial hot water, which they inevitably will.

Staking your entire business on the health and goodwill of another business you have no control over, as many tech companies are now, seems astonishingly shortsighted to me. But everyone's doing it.
posted by drjimmy11 at 12:31 PM on May 4, 2013 [2 favorites]


I wonder how many companies would offer free services if regulators prevented them from shutting down those services.

If Google had never provided free Reader in the first place, any one of the small independent services they displaced might have gone on to grow a sustainable business. Regulation of this sort of thing is obviously a minefield, but maybe there’s a connecting path to EU-style personal data ownership law or predatory pricing law. We’ve been doing this long enough with enough examples of the perils of free services that it should be possible to discuss some form of legal structure. It’s time for the big boys to grow up.

I just don't like seeing other people get fucked.

+1 to this.
posted by migurski at 12:34 PM on May 4, 2013 [3 favorites]


Having re-read my own comment, you'd think it would be standard practice for sites to do some kind of failover like

if (cloudUnavailable) useLocalAssetInstead

But I've never seen anyone do that or even consider the ways in which they might do that in an emergency, or really give any thought at all to the fact that the "cloud" might someday not be there.

There are a ton of major sites right now where, if Amazon S3 has a problem, the site just goes down until Amazon fixes it. It's a pretty amazing step backward as far as reliability, but no one's saying anything.
posted by drjimmy11 at 12:35 PM on May 4, 2013 [2 favorites]


Gwern's whole website is just so great. I am in awe.
posted by vanar sena at 12:44 PM on May 4, 2013 [1 favorite]


But I've never seen anyone do that or even consider the ways in which they might do that in an emergency, or really give any thought at all to the fact that the "cloud" might someday not be there.

The Obama campaign used failovers for services like payments, more along the lines of “if (cloudUnavailable) useOtherCloud”. Google’s Cloud Store is an API clone of S3, so it’s possible to use them as backups for each other. Many services such as edge caches, cloud computing, relational data stores and others are commodities, so you can swap Fastly for Cloudfront or Akamai, MySQL on hardware for RDS, hosted Memcache for Elasticache, etc.

Heroku is one that I fear, though. There’s not an equivalent competitor for much of what they do, and I’ve heard that they’re located exclusively in the AWS us-east-1 data center. Git is decentralized by design though GitHub is not.
posted by migurski at 12:46 PM on May 4, 2013 [1 favorite]


JHarris: "not everyone has the resources to run TinyTinyRSS, nor the technical know-how to set the darn thing up."

I need to write a blog about this or something, but you can pretty much run a personal webapp site for free these days. Redhat's OpenShift, for example, has a free tier, and people have already written the deploy scripts for you. Heroku has a free tier, and EC2 offers a free micro for a year (per account). I pay 5 dollars a year for DNS, but you could use a dyndns service for free.

I'll let you decide whether free Platform as a Service is going to be more resilient than free Google Reader. But at least they have a paid model in place, and you have full access to the entire data.
posted by pwnguin at 12:49 PM on May 4, 2013 [3 favorites]


Never long in these threads before folks show up to tell us that not only should companies be able to do whatever they damn well please, they should also be able to do it completely free from any criticism at all, huh

Gosh, those entirely imaginary people that you're inventing out of whole cloth are such assholes!
posted by yoink at 1:24 PM on May 4, 2013 [1 favorite]


> You can't really predict this stuff from usage stats or even revenue.

w0mbat, I disagree. I'm not saying that you can predict this stuff with extreme accuracy, or that a huge amount of the variance in lifespan is being explained by the covariates I specify. But I think there is non-zero contribution there. (And really, given that most people have not the slightest idea how long a Google stuff might last in the first place, even if the covariates were completely useless, the data would still be useful just to tell what the data *looks* like.)

mwhybark: Not quite :) I only just registered today so I could reply to comments here. ('Someone is wrong on the Internet!' I'm sure everyone here understands...)

vanar sena: Thanks.
posted by gwern at 1:31 PM on May 4, 2013


gwern: Well then you haven't exported your data, have you? Admittedly, if all you want to keep is the copies of those disappeared web pages then you could just print to PDF.

twoleftfeet: And that is why I use a bookmark based RSS reader for firefox. There have been a few version issues, but there has always been a replacement (Sage, then Sage Two, then Sage again, and some other one between Sage and Sage Two that I forget. Anyway, if I really needed apparently editing add ones to fix incompatibility is easy if it isn't actually incompatible, it just claims to be.

whir: To be honest, I've not used Google Reader in years, as it felt like a poor man's Harbi Xenu/Sage (Both Firefox extensions). I heard a bunch of people whining about it going away and assumed they'd lost something important. If they can just export everything, why not put it into another feed reader? It sounds like there is some data that you can't export.

meowzilla: It is the job of goverment to force companies to do things they don't want to do, but should do. For example, forcing them to make sure products are safe, pay workers ethical wages, etc. People seem to forget this a lot.

JHarris: Sage. Firefox extension. Works much better as you can see the article headline before clicking it.

zabuni: You could make an exception if the company was closing down: There is a difference between a company closing down, and deciding to flick a switch or be bought out. Also a year seems really long to me. A month or six weeks should be more then enough for users to get their data out and find a compatible service.

tommasz: Gmail offers both IMAP and POP3. I have a full backup of my email on my laptop.

davejay: You see that as a bad thing. I see that as forcing companies to do realistic assessments of their ongoing costs, and that would likely push people away from this horrible idea of cloud based services for everything, and back to reliable, transferable installed software that has none of these issues.
posted by Canageek at 1:33 PM on May 4, 2013 [1 favorite]


pwnguin: Or you could, you know, run it as a local program thus avoiding all web hosting costs. I really don't get why you would bother to run it online when all it needs to do is pull RSS feeds off the internet and parse them.
posted by Canageek at 1:37 PM on May 4, 2013


We are told that criticism of Google is unwarranted because we just don't understand how free markets work.

Wonder what the same folks think of Walmart.
posted by dhartung at 1:50 PM on May 4, 2013 [4 favorites]


If people seriously do this then they should go to extra hell. Offering a product, even for free, implies that it's going to have some degree of continuity. . ... I will never give them so much credit again.


Apple ][ forever owners, Newtons still being an important education offering the day after being cancelled, and listeners of "plays forever" would like to introduce you to the "Beat you like a piniata with a clue by 4" support group to drive home the idea that nothing made by man or things that pretend to be man lasts forever.

If you want to have your software based do-dad last as long as you want to use it, best go open source that you run on your own hardware.
posted by rough ashlar at 2:18 PM on May 4, 2013


> gwern: Well then you haven't exported your data, have you?

Well, I suppose if you define 'export your data' that way...

> Admittedly, if all you want to keep is the copies of those disappeared web pages then you could just print to PDF.

My understanding, from a guy who used to pull full site histories out of Reader, is that Reader has added some sharp cutoffs and difficulties to doing so. You may be able to find pages through search but you cannot pull all of them out; this is a problem since generally you do not and cannot know in advance exactly the pages you will later need...
posted by gwern at 2:25 PM on May 4, 2013 [1 favorite]


Canageek: "pwnguin: Or you could, you know, run it as a local program thus avoiding all web hosting costs. I really don't get why you would bother to run it online when all it needs to do is pull RSS feeds off the internet and parse them."

As I understand it, the value here is being able to sync across devices. The difference is like POP3 vs IMAP. I can kinda see the value there, and I might try out ttrss integration with Liferea.
posted by pwnguin at 3:29 PM on May 4, 2013


Offering a service implies it'll be around for some time;

Nonsense. This is simply not remotely true and no one has ever thought any such thing. No one even think that about "free services that are offered on the internet" as a general proposition--and god knows people have a bizarre sense of entitlement about things that are offered to them on the internet.

The internet is simply littered with tools that last for a little while and then die. Nifty little insta-calculators of one kind or another, or apps that you can pop onto a webpage that give you stock quotes or the current time/temperature in a bunch of places, or whatever. In the vast majority of cases our response to seeing them die is to shrug out shoulders and say "oh, that's a shame, I really like that thing; I guess whoever made it couldn't figure out a way to make it pay" or "I guess that grad student moved on and lost interest in that side project."

What makes people angry about Google dropping services is not an implicit promise that "any service ever offered, even for free, will be offered for ever and ever amen." What pisses people off is the fact that the service was, in fact, "around for some time." They came to rely upon it; they came to take for granted that it would always be there. Had Google Reader lasted for a week before they pulled the plug on it, no one would give a damn about it one way or the other. The irony here is that the bigger the favor Google do you, the longer they give you something you find useful for free, and the more useful you find that free thing, the more pissed off with them you're going to be if they ever cease to offer it. But the notion that offering something for free for a long time entails a moral obligation to keep offering it for free is simply ridiculous. If you agree to run your friend to the airport every time he asks for two years, does that mean you have ceased to have the right to say "no, sorry, it's really not convenient"?

Now, there is, of course, a separate argument about what is commercially wise for Google to do. If they want people to regard them as a reliable company, is it a good idea to piss people off? But that's a question about what is the best business strategy for Google to adopt; it has nothing at all to do with whether they are under some kind of moral obligation to continue to provide services for free in perpetuity.
posted by yoink at 5:04 PM on May 4, 2013 [1 favorite]


Yay for the Kaplan-Meier estimates! So glad to see someone actually using survival analysis to talk about this, rather than ignoring all the still-surviving Google services.
posted by mixing at 5:24 PM on May 4, 2013


Someone should make an activist group to fight these shutdowns.
posted by jscott at 5:51 PM on May 4, 2013 [4 favorites]


Jscott, any plans to release a version of warrior that can be installed on a base Linux from Git + scripts?
posted by migurski at 5:57 PM on May 4, 2013


Well, we have the "advanced" version but I generally like people to come hang in the IRC channels when we pass those to people, since if things go south, they go REALLY south with someone doing their own thing.
posted by jscott at 6:03 PM on May 4, 2013


gwern: By export your data I mean, everything you have saved in it is able to be copied out and in a format another program can open (ie well documented)

pwnguin: Isn't that exactly what Firefox Sync is for? That is how such services should be done: It is on my computer, and every computer I sync it with. If Mozilla and all its computers vanished tomorrow, my copy of Firefox would still work, and have its bookmarks and extensions (Though I wouldn't be able to sync them anymore). Other things that do this right: Mercurial and Git. All the code is on my computer. It is also on Bitbucket (or github). I can put this code on any other computer easily, but if github or bitbucket go offline, I don't lose anything, and I can easily set up my repositories to link to the competitor with a trivial amount of effort.
posted by Canageek at 6:14 PM on May 4, 2013


I'm surprised Toolbar made the "least risky" list, given that they already killed off Toolbar for Firefox.
posted by IndigoRain at 7:11 PM on May 4, 2013


>Offering a service implies it'll be around for some time;
Nonsense. This is simply not remotely true and no one has ever thought any such thing.


Then why are so many of us so angry about Reader?

The fact is, most of the services, if they're not going to be around for some significant period of time, then there is little point to using them.

Canageek: I've heard of Sage, but does it let me back up feed lists and articles? Does it sync between Firefox installs?

I remember back at GSU frequently loading up Reader on computer lab machines and keeping up with my sites. I'm no longer at university but I'm still saddened this option is no longer available.
posted by JHarris at 7:20 PM on May 4, 2013


I looked up the services listed in the deadpool (ahem, I mean "most at risk products"). Boutiques, Magnifier and Hotpot all appear to be dead already or merged into other products. Unofficial Guides must be called something else because I have no idea what that is or how to find it (using Google, natch).
posted by chrominance at 8:02 AM on May 5, 2013


chrominance: Unofficial Guides is a series of books publisher Google bought (yes, I know - who knew Google had bought some book publishers?) and I think they're either still publishing them or they got sold back to the founder or something.

> or merged into other products

As I specified in my criteria at the very beginning of data collection, merger counted as still alive. Who cares about the branding as long as your functionality is still there?
posted by gwern at 10:37 AM on May 5, 2013


JHarris: What is a feed list? Also: Just bookmark article if you want to save it Or the Wayback Machine version if you want to be sure. Or print it to PDF. It works on Live Bookmarks (Bookmarks to RSS feeds) so all you have to do is set up Firefox Sync and yes it will then synchronize across Firefox installs.
posted by Canageek at 2:25 PM on May 5, 2013 [1 favorite]


Voice shutting down would be kind of interesting, as mine is actually a Sprint integration with Voice. Annoying, too, since I do like the fact that texts and voicemails get forwarded to my email.
posted by tavella at 2:31 PM on May 5, 2013


JHarris: What is a feed list?

It's not just a collection of feeds, Reader is an archive of every RSS entry a feed emits since it began, in a simple, no-nonsense list. If it disappears from the feed file Reader will remember it, and let you scroll backwards through it, in chunks, all the way back to the beginning. That is awesome, and ensures that you don't miss anything in the feed, meaning you can afford not to load Reader obsessively; if you don't come back to it for a week everything you missed will still be there. Even if the site makes posts accidentally and deletes them, they'll remain in Reader. I don't think any of the Reader alternatives can match that yet.

Bookmaking things to save them? Really? Let me introduce you to my overloaded bookmark list, half the links on them having died years ago. Saving each one, one at a time, as PDFs? Let me laugh in your face! Relying on Wayback's extremely slow server to fetch the page, and hope its spider got all its images? If you tried using it for that purpose you'd know how woefully inadequate that process is. The Internet Archive does good work and tries very hard, but lots of stuff falls through the cracks, a fact I've learned all too well.

Reader is as popular as it is, and Google as lambasted as they are for shuttering it, because there is no adequate alternative. It was even better before they killed sharing those bastards.
posted by JHarris at 2:54 PM on May 5, 2013 [2 favorites]


It'd be nice if you could be more specific in your criticisms.

My criticism is that the entire exercise is GIGO. Even if product cancellations could be modelled the variable this guy is able to masure are unlikely to be the ones that drive the decisions. Issues like technical debt, engineering resources and internal politics are all more likely drivers than whether something os open-sourced or whether the service is "profitable." And I say that in quotes because the way Google does internal accounting is unlikely to be clearly obvious to those outside the company. Is Android "profitable"? What revenues are or aren't counted towards Android? There's a definite answer, but if you're not one of a few dozen googlers who know it then you're just guessing which leads to GIGO analysis.
posted by GuyZero at 11:04 AM on May 6, 2013


> My criticism is that the entire exercise is GIGO.

You're conflating multiple aspects here.

1. Even if the covariates prove to be completely useless, the analysis still has predictive value because no one knows what the overall Google population looks like. Quick, what's the median lifetime?! Oh, you don't know? Well, I don't blame you, I had no idea either. In a state of such ignorance, merely describing the population lets you make more accurate predictions than you did before.
2. The covariates didn't turn out to be useless.

> Issues like technical debt, engineering resources and internal politics are all more likely drivers than whether something os open-sourced or whether the service is "profitable."

All these things contribute to variation, yes, but we have a thing called 'statistics' for dealing with such randomness. You don't need to know exactly how the black box works to improve your predictions.

And things like 'engineering resources' don't matter. Of course resources are limited. There's always a limit, that's why Google has not consumed the earth and become a ball of dense servers expanding outwards at the speed of light. Imagine all the products are standing in a pool and cannot swim; if the waterline goes above their head, they drown - we are interested both in tallying how many swimmers have drowned every summer and also in measuring their relative heights to know who will drown first. We don't need to know the exact volume of water down to the milliliter.

> Is Android "profitable"? What revenues are or aren't counted towards Android? There's a definite answer, but if you're not one of a few dozen googlers who know it then you're just guessing which leads to GIGO analysis.

And there could be a teapot in orbit around Mercury, who knows? I guess it's all GIGO!
posted by gwern at 12:53 PM on May 6, 2013 [1 favorite]


Metafilter: a ball of dense servers expanding outwards at the speed of light.

*cough* I think you forgot a "yet" on the dense ball of severs expanding outwards. I'm sorry, I'm not the only one seeing links between Eclipse Phase and Google?

Also, I now want to launch a teapot into orbit around Mercury just to screw with both of you.
posted by Canageek at 10:04 AM on May 7, 2013



Also, I now want to launch a teapot into orbit around Mercury just to screw with both of you.

Will a huge boiling teakettle orbiting Saturn do the trick?
posted by jamjam at 12:43 PM on May 7, 2013 [1 favorite]


Looks like an important chunk of Google Code is now going away.
posted by Blazecock Pileon at 4:16 AM on May 23, 2013


Yes, but Github did the same thing months ago. It seems like there might be an abuse of some manner. Copyright violation, perhaps?
posted by pwnguin at 12:19 PM on May 25, 2013 [1 favorite]


They also disabled uploads for mailing lists/Google Groups a while ago too.
posted by gwern at 2:32 PM on May 25, 2013


« Older "Millions make me lose interest. It should have a...   |   Three Ohio Bucks Found Drowned With Antlers Locked... Newer »


This thread has been archived and is closed to new comments