Hacker News new | past | comments | ask | show | jobs | submit login
Google Killing Off Buzz and Code Search (googleblog.blogspot.com)
318 points by thisisblurry on Oct 14, 2011 | hide | past | web | favorite | 144 comments



So, in 2005 I was involved in a company that thought "man, it would be great to build a search engine for source code". I even started putting together components for it, such as a model for finding text inside of larger documents that had statistical properties similar to code, auto-detecting the language, so you could find code snippets inside of blog posts.

However, by the time we got organized enough to actually do it, Google Code launched, and had this really awesome code searching feature, that everyone considered to be "more than good enough" and "comprehensive, as Google is indexing the hell out of stuff like this".

But, now, Google has now determined that that wasn't sustainable, and has shut down the project. Which means that both our company, and any other company, that thought it had a sustainable model for running such a project, and at this point would probably be "pretty awesome", never started, and we are all suddenly thrust back into 2005, unable to search for code.

This... (I now emphatically point at the previous paragraphs) is why I don't like Google very much: they have such large resources available to them that they tend to just swoop in and offer an unsustainable service at a loss, training users that "things should be so free it hurts: in fact, they must be losing money on every use of this" (Google Voice being a great example), thereby stifling innovation by people who can't possibly undercut that.

Note: this isn't even a problem specific to Google... startup companies that get VC money tend to also cause this problem. They get tons of money, offer a service at a heavy loss while they use that burn time to determine a business model, actively knowing that they are operating at a loss in order to get users as fast as possible from other people who might try to get them.

Of course, the result is that the company usually either totally implodes (typical of any startup) or, even more insidiously (for the projects that actually becoming successful, even quite popular / common), come up with a business model so ludicrous that the users actively revolt against the entire concept of the service...

... and, where do they go? To some other free service offered by another company that managed to get equally large sums of VC money because they point at that other company that had hundreds of millions of users that just failed because of a bad business model, something they will know how to fix (in a couple years or so, once they get around to figuring that part out...).

:( I liked Google Code search, and I'm going to miss it.


Which means that both our company, and any other company, that thought it had a sustainable model for running such a project, and at this point would probably be "pretty awesome", never started, and we are all suddenly thrust back into 2005, unable to search for code.

I see two viable reactions to this:

1. There's now a vacuum where Code Search was, and, evidently, some demand. Why not startup now?

2. Not even Google could find a viable model for Code Search, even with their resources and scale, and (ugh) synergy. Google did the research for me. (And if you think Google dropped it because they couldn't find a viable free model, and you think there's a viable paid model, see #1.)

As for startups, the phenomenon you describe seems to apply mostly to things that depend on network effect and probably advertising. If you have an idea that brings value to someone and can demonstrate a sustainable business model based on charging for that value, I would imagine that to be a much more attractive option for people who want something they can depend on than a VC funded free product startup.


1) I am not whining as an entrepreneur that I did not start the company in 2005: I did other things that I am quite proud of, and which frankly have probably been more interesting and successful than having worked on that code searching mechanism.

Instead, I am whining as a user of Google Code search that we have just lost six years of progress on searching source code. Having a company that starts today in this space, while they will have an easier time (due to advances) starting and getting up quickly, is drastically different than a company that started six years ago, and might have a large index, advanced algorithms, etc. at this point.

2) I don't think Google even tries to make things sustainable. During all of these "Google shuts down X" that have been happened recently, some posts here by ex-Google people indicated that Google internally didn't even have reporting on per-project costs... hopefully now that they see how much certain things cost they will cause less market-level problems going forward.

(edit: I forgot to add this point) I mean, seriously: Google, a company that makes all of its money on advertisements, and for whom ads permeate everything, doesn't have ads on Google Code Search, and AFAIR never did. If you do absolutely nothing else for such a service, as Google, you'd put some ads on it; even Gmail has ads, and that's a place where people find them creepy.

I therefore simply cannot believe that they actually are looking at this project as something that is worth their time to make sustainable: they are a giant company with a thousand things to do; this was cool one day, maybe useful to some engineers internally, and they kind of just did it, and dumped the result on the ecosystem, where it created a little vortex-of-free-and-good that eventually has collapsed into a singularity, and will be gone in the next couple months, leaving us with the "vacuum" that you mention.

Meanwhile, it might simply be because even if it was successful, the userbase would be tiny, and the revenue generated would be "not enough to justify the lack of focus" (Google is getting behind the whole "more wood, fewer arrows" thing lately); whereas, to a small company, it might have enough margins to sustain a few people, which is "enough to bother with" at that level.

~3) The point here isn't that people who care about sustainability should be using Pinboard.in instead of del.icio.us (although they should, and those that didn't got burned at the last minute): it is that society is worse off because there are people in the ecosystem who are operating unsustainably.


I see your point. It's a good one, and I sort of agree with you, but I also think it's more complicated than that.

For example, it's hard to say whether the six years of code search we got from google was better or worse than what a sustainable startup could have provided. And if was better, how do you measure that against its sudden disappearance until there's some viable replacement (if there will ever be one). It's also hard to say whether it was ever possible to run it sustainably.

I think for every code search, there's an equally unsustainable (at first) Google product that provides a lot of value for free to users.

Yes, sometimes it means that they kill opportunities for others who could possible make sustainable businesses, but it also means that they can run something unsustainably for some period of time to see whether it's worth it to even make sustainable.

It's hard to say whether overall for society there have been more cons than pros. Anecdotally, my experience is that Google has provided me a lot of value for free.


I am not certain if you saw the edit I added to my previous comment (I added it reasonably immediately, but you also responded within 10 minutes, so maybe you missed it), but I believe it addresses some of these points; I will restate, however, more directly targeting this post, just in case.

Google doesn't seem to try to make things sustainably, and in this case didn't even try their bread-and-butter business model (ads), possibly because to them the result would have been "chump change", whereas to a smaller company it might have been "enough to man a few bright people" (which I'd argue is enough to have something great occur).

I'll add, though, that these services Google (or the VCs whose money is being used in an unsustained burn) is providing "for free" are coming from somewhere: from other projects, if nothing else. In the end, everything has a cost, "there is no free lunch", yadda yadda.

Regardless, point taken that as a user who only cares about that one specific feature (which I admit my arguments skewed towards), it might be the situation where a large company burning money into a pit for six years may actually cause more money to be thrown at a small problem over the lifetime of that problem than a sustainable market, and therefore we might be glad in some twisted way about that. ;P

I still feel, however, that from a market-level, projects like this are bad for everyone, in the same extrinsic way that we look at people using seemingly cheap materials like styrofoam and partially-hydrogenated oils to build empires that eventually have seriously bad costs; put simply, I look at Google and VC-backed startups in the same way people look at companies polluting the atmosphere and contributing to global warming. (and yes: this is now pretty abstract, but I feel this conversation started getting to a position where the abstract-ness was fun.)


I enjoyed reading your back and forth in this thread. I can see both of your points and don't really know what side of the fence I land on my self but I would like to jump in with one observation you might find poignant.

To carry on with your climate analogy... My economics professor used to talk about "the destructive winds of innovation or invention". I think this might be a perfect example of this. Markets are a messy place where new ideas displace old ones, new models challenge old models and sometime you compete with companies that behave in unsustainable and irrational ways. I think you can wish it wasn't that way but I tend to believe this is a symptom of the human condition. What alternative is there? Surly none of us want the government to regulate what projects companies choose to work on? We are stuck with people self regulating themselves and this always has mixed results.


Just because it is not rational to assume that people will not act unsustainable, does not mean you have to like the people or companies that do. Just because I know that there will always be someone polluting the ecosystem to make a profit, doesn't mean I should shrug my shoulders and not write articles about why that's a bad thing for other people.


Wow, that's actually a very interesting position, and very interesting metaphor. I'll have to think about it before I can really react, though.

Unrelatedly, this thread has been an absolute pleasure. Thanks for not dismissing me for the Google fanboy that I am.

Also, it's kind of awesome to be conversing with you on HN. I've used your jailbreaks and software on iPod Touches before I moved to Android, and I think it's awesome work you're doing. And is there a place where you've talked about your reasons for working on iOS instead of a more open platform? I'd love to read your position on that.

Thanks!


I only know of one reasonably open cell phone at the time the iPhone came out: Open Moko (these people seriously had scroll bars in their mobile UI, and their hardware was clunky, expensive, and unreliable). Some of the Palm and Windows Mobile devices had been hacked (a friend of mine managed a port of Debian to the iPaq, "Familiar"), but I do not remember any of these devices having "you are allowed to reflash us" as a feature.

After the iPhone came out, I got involved with the G1 for a while. I made a few code contributions to Android; however, most of my patches, even with one of them being a possible security issue, took six months to get into AOSP... it wasn't a really open process. By the time they got back to me I had forgotten what I had learned about the things I was fixing, and never really looked back on the idea of contributing more.

Additionally, the G1 wasn't actually an open hardware platform: we had to hack it. Someone figured out the "telnetd is installed and the keyboard is attached to an open root console" trick, but then the "real work" started getting a bootloader exploit; it was on a mailing list I ran (g1-hackers) where the bootloader was first dumped (possibly even on my phone, actually, although I did not do the hard part: I just did the labor; I feel like the guy who did it was named Edward).

The "open" Android device, the ADP1, was expensive (you couldn't get it subsidized, only unlocked), and frankly: it was a G1... the G1 was not a good phone. If you had an iPhone, especially one you already had root access on, the Android devices at the time simply were not interesting.

Since then, Android has become more open in some ways (more pieces of hardware from better manufacturers allowing you to unlock the bootloader), while more closed in others (the source code becoming more and more locked down: right now 3.x is still a closed branch; even when it opens, Google has promised longer delays between releases to AOSP).

Meanwhile, Android's quality has continued to trail the iPhone: despite sometimes even having a faster CPU, the fact that it has a low quality JIT backing the entire system causes the experience of the phone to be "slow"; worse, it is incredibly laggy due to the high latency IPC that is used to interface rendering things in Java to the incoming touch events... until 3.x it didn't even have hardware accelerated UI compositing.

In comparison, despite being "closed", thanks to the work of people working on jailbreaks, I have always had a playground where "the phone is mine"; and, thanks to the work that I do with MobileSubstrate and Cycript, I've never actually felt that limited by the fact that the source is closed, as I go to make modifications to things: I just do it anyway; frankly, Objective-C ARM binaries are very easy to read ;P.

In fact, while looking at Android, I feel like the key problem with the people who work on hacking the ecosystem is that they have become reliant on the source code being available, which has not only led to work slamming to a halt while 3.x is behind closed doors, but also has meant a poor experience for users: people have to install entire replacement ROMs (compiled from AOSP), rather than just installing the specific modifications they want (as people do from Cydia).

Finally, we can look at "what hardware would I use if I could install either operating system on the device", and the answer right now would be the iPhone 4S, and even without that device (as I have no jailbreak for it currently), the answer would be the iPhone 4: the reason is, primarily, the high resolution and high quality screen.

It is for this reason that I am anxiously awaiting the Nexus Prime, which I have heard will actually have a reasonable screen: supposedly, it will be 1080p, which is even a higher resolution than that on the iPhone 4 (but watch the entire device be ludicrously larger). Unfortunately, (for what I think was probably a good reason,) its launch has been delayed for a bit; we will soon see.

It will also, supposedly, not be Pentile, which makes the Nexus S that I carry around with me even worse than the Droid I had previously. However, I hear that it will still be AMOLED, which has a really really bad problem with burn-in (on my Nexus S, just a few hours at the lock screen is enough to have a noticeable shadowing where the unlock slider is; even for normal users, the status bar is up constantly).

So, I guess those are my reasons: 1) history (iPhone was the first to market), 2) equality (frankly, these devices are all closed), 3) quality (the iPhone hardware and software continues to be hands down the best). That said, I am doing a lot more on Android recently, as its devices and operating system have started to become reasonable, but certainly not because I consider Android to be "open".

And with that, I will leave you with a failing command (and yes, I realize that kernel.org was hacked, but Android didn't even care about finding a replacement), straight from the Android lead himself. (Note: this definition doesn't even capture the key "make install" step that Android devices are hit and miss at; and don't even get me started about how the result isn't even a complete system.)

@Arubin: the definition of open: "mkdir android ; cd android ; repo init -u git://android.git.kernel.org/platform/manifest.git ; repo sync ; make"

-- http://twitter.com/#!/Arubin/status/27808662429


Given your bent on sustainable business models, Silicon Valley startups, VCs, and the effects of the existence of Google-like entities, you might find interesting Jaron Lanier's recent brain dump[1] on those topics and a little more. It's a one hour video, and the most immediately relevant topics are sort of far along, but there's a (cleaned-up, prosified) transcript available, too. (That might be a bonus in its own right, if you're someone who prefers to read rather than waiting on a speaker's own cadence and pauses.)

1. http://edge.org/conversation/the-local-global-flip


I think google code was very expensive to run because of the regular expression search feature, which was awesome, but cpu intensive... I'm sure they are killing the project because of feautres like that.


If that was the problem, they could have simply eliminated that feature and kept a simple substring search.


Reminds me of this story. Substitute Apple for Google and Panic for you:

http://www.panic.com/extras/audionstory/

Excerpts:

We [at Panic] also seem to remember that Jobs painted us a vibrant (but genuinely honest) picture of how he viewed Audion [our product] fairing against iTunes:

"It's like you guys are a little push-cart going down the railroad tracks, and we're a giant steam engine about to run you down."

  ***
I remember [Jobs] asked, "Do you have any other ideas for apps you want to work on?"

I replied, genuinely, "Well, we've got an idea for a digital photo management program..." and he replied with a simple, "Yeah. Don't do that one."

Everyone in the room laughed but I had no idea why — remember, my head was still exploding — so Steven Frank had to explain to me that he meant, basically, it was already being made and, of course, it would be called iPhoto. Oh. I get it now.


Thanks for the link. Very educational. I see it was discussed on HN long time ago.

However Audion story is quite different from Google Buzz / Code Search story. Apple didn't kill iTunes. Audion would have died anyway, because they didn't find free business model for their MP3 player.

Audion story is educational in terms of missing key business opportunities.


Similar situation in 2007.

If anyone had any doubts, Google is the new MS. In the mid 90s, any time I talked to people with ideas for new ventures, the inevitable "how will MS play in that market?" question came up, and often killed ideas before or just as they were getting started. People had seen MS buy up companies then shelve things, or enter a market (whether they should have or not) and effectively take the competition out of it - no one wanted to compete with MS. Maybe they could get bought out, but competing would be hard, and investors would think twice before investing against MS. To the degree that it validated a market or idea, that was good, but they were too big for people to think they had a shot at competing.

Few people are afraid of MS, but Google seems to have replaced MS in that role - intentionally or not. When your goal is "organize information", almost any endeavour could fall under that heading. I don't think people are necessarily as afraid of Google as people were of MS 15-20 years ago, but the effect on many ideas - like saurik's - is the same outcome.


I think it's worse with Google, honestly. It might be that I simply wasn't able to analyze Microsoft much during their heyday (generally too young), and I'm not saying it is necessarily /much/ worse, but at least in the theory of this model I'm using, it certainly seems worse; like, I blame Google more for their actions than I do for Microsoft.

Microsoft wanted to win, and in so doing was brutal in their attempts to control things and take over segments of new markets; and when things do fail, they either doubled down or shut them down pretty quickly. They were a fierce competitor, but they actually wanted to make money.

Google, on the other hand, seems to just stumble around, kind of clumsily and seemingly even accidentally tripping over entire markets. At no point do they attempt to make money, and their involvement often seems trite: releasing something, and then never updating it again.

It's like no one at Google ever cared about making money with the new thing they launched, and sometimes they don't even seem to care if it "succeeds" or "fails": if you talk to them they kind of shrug the notion off as if they just wanted to build it, or maybe they needed it for Google internally and decided to let the world benefit from it.

The result is a ton of stuff that is nigh unto intentionally poorly maintained that, due to having access to Google's index, or Google's infrastructure, or simply Google's warchest, manages to be "the king of the market", despite any other shortcomings it has.

In fact, the only situation I can come up with quickly where this happened with Microsoft was Internet Explorer 6; where, for whatever reason, they seem to just stop working on the project. (You could then argue that, retrospectively, IE3-5 was "part of the problem" for being free, but I think that was actually "mostly ok"; I'd find the argument interesting, though.)

In comparison, Google has tons of projects (or at least did, before this year), where I've seen startups with "better stuff" get devoured by something for which I don't even see an attempt at a business model from Google (and then which they often never update again).

My favorite example is actually Google Voice, because some of the competition in that space is so good that they can actually convince people to buy it even though Google Voice is free; some have better transcription, some have better group features, and some are simply easier to use.

However, the price points are whack, as are the economies of scale; each of these players is playing in a tiny little market that almost no one will ever hear of, because their competitor (Google) is seriously just bleeding money: giving people phone numbers and forwarding arbitrarily long calls, for free.

(edit:)

Thinking about it more, Wave did seem to me quite a bit like Microsoft: a ton of companies I had seen that previously were looking at doing collaborative things heard that Google was releasing Wave "someday", decided "Google will win" and that to survice they needed to "interop", in so doing got involved in a beta program that lasted seemingly forever, and after it was actually on the market for a rather short period of time, Google scrapped the whole thing. <- Classic Microsoft

But the effects of that were then only "a couple years", and during those years Google threw tons of marketing and engineering effort at the project; when Google did decide to exit, they did so rapidly: there was no "extended sunset period" where Wave continued to dominate "by accident". Wasteful? Yes. Annoying? Yes. Google Code or Voice level of unsustainable? No.


I think in the beginning MS wanted to win, but I think eventually they got to the point where they got in to markets just to establish some market presence, letting competitors know that they were interested in playing in that space, and preventing others from making too much headway in that space.

I don't think Google's forays in to myriad markets is necessarily motivated by any real interest in being in specific markets - I think often its a progression from their 'labs' idea. Arguably this is less aggressive than the MS of old, but... I think the effects are worse in some ways, as you see with some projects. Code Search was one. Google Voice may be another.

Loads of people end up flocking to these services to the detriment of smaller, focused competitors. In many cases, it's probably because the services are free, but also Google's "aw shucks, we're just being good guys" wins a lot of people over. Google has an insane amount of power/control over peoples' personal data and activities compared to the MS of 15 years ago, but people trust them similarly to how people viewed MS in the early 90s - Windows 3.1 was great, and Win95 took the world by storm.

The game is Google's to lose right now, and I wonder when that happens if it'll be gradual, or instant.

Good point on the Wave plug pulling. I suspect though that if they had even a fraction of the paying customers MS did for any even minor MS product, that they'd end up with a long sunset period as well. Google generally doesn't charge for many services, so I think people cut them some slack on the fast shuttering of those same services. However, it's generally the case that most people have quit using it (and, again, they're not paying anything for it anyway - here's your money back). Google killing 'Buzz' is another example of that.

What's missed in notions of 'attention economy' ('ads') is that Google monetizes our attention. When they close a service, we can't get that attention back, but they do get to keep the money they earned from our attention. In a more traditional MS-style product cancellation, there might be a chance of a refund, or we'd at least get to keep the product.


"I suspect though that if they had even a fraction of the paying customers MS did for any even minor MS product, that they'd end up with a long sunset period as well."

I'm sorry, I must have been unclear: it has been my experience that MS /does/ pull the plug swiftly and unambiguously when they do intend to give up; they "double down or shut it down", as I tried to describe in my post. It is Google that I find tends to just give up on something, but leave it there, as if it still works, as if it will always work, and goes on to do other things while it rots.


I see your point rereading (thanks for clarification) but it's always felt to me like MS has left things hanging around as well, but I can't think of one (beyond IE6!) right now :)


I agree that when price of zero is bad in sense that it raises the users expectations to infinity (if you think of value/price) and this is bad for the echo system. On the other hand "free" allows everyone in the world with Internet access to use the service which is a very positive thing overal. IMO, the best model (in terms of impact, not in terms of business model necessarily) is a hybrid where you offer premium services (like github, linkedin, dropbox, etc.). I'm not sure why Google shies away from it but my guess is customer service.


I actually agree with the point of your comment: GitHub, to take one of your examples, seems to be a profitable business model. My reaction in this context, however, is that I do not feel like Google is trying to figure out a business model of any kind for most of its products.

To try to draw the difference, in the case of where Google offers free things, it seems to not have any relation at all to the things where it makes money: I don't think that by offering Google Code Search they got more people using Gmail or Google Search (things they monetize).

The best idea I've seen so far is that by doing large numbers of totally random fun/free things "for the world", users will always see them as "not evil", even though their core business model (aggregate information on users and use it to drive targeted advertisements) is the same as that of companies like DoubleClick (generally considered "evil").

(Which I found to be a hilariously awesome notion: that Google's random projects /are/ "freemium" in a sense, but where the upsell is more just a general notion of trying to dampen an overall negative connotation. As Google shuts down side projects, and concentrates on more controversial things like Google+, it will be interesting to see if they manage to maintain their "people don't think we are evil" status.)

(To try to be very clear: I do not consider advertising to be intrinsically evil, and am even sometimes involved in that space: do not take that previous comments to be indicative of a reaction I personally am having to their business models. Lots of people hated DoubleClick, lots of people seem to hate most ad companies, and yet people generally like Google: I think trying to analyze why is very interesting.)

...

Now, that aside, I think that there is still a bug (but find this to me more on the philosophical side): you can also view the freemium business model as a "tax" (subsidizing a general service by taking a cut of another area), and whether that tax makes sense or not becomes tricky. In the case of GitHub, their pricing for people who want private repositories is brutal, and those people are subsidizing the ecosystem of free projects that are using GitHub to host their code.

This, however, is not actually "sustainable": people on HN constantly complain about the cost of using GitHub for private repositories, and lament that there isn't something at GitHub's level competing with them that they can switch to, which won't cost them hundreds of dollars a month for what amounts to "almost no disk space or usage" (as GitHub bills per repository).

The problem here is that if you have a large free service that you are using to bootstrap your paid service, the paid service is in danger of being attacked, and being attacked hard: if nothing else, people who are offering the paid service competing with you aren't distracted by the free service (I am brutally aware of this bug, as my actual business has this problem; it is a little better for me, though: my market position is more similar to if GitHub also was the company that primarily developed git itself).

So, last week, BitBucket announced git support, and has what many people here consider to be reasonable pricing. I am now very curious how long it will take before people currently paying GitHub take strong enough notice (maybe requiring upgrades to BitBucket, delaying the process further), and start to switch.

When this happens, GitHub is no longer going to be able to extract the tax it currently extracts, and is going to be in the interesting position of having a large number of free users it is trying to support: people who now consider social version control to be something that is intrinsically free, and which large amounts of the open source ecosystem now rely on.

If that happens, it will be quite interesting to see how it goes down, and who gets effected in the market and ecosystem implosion that results.


> In the case of GitHub, their pricing for people who want private repositories is brutal

Well, I wouldn't call $7 (5 private repos, 1 private, oo public collaborators) to $20 a month (10 private repos, oo private collaborators, oo private collaborators) "brutal".


I used it as an example as I know HN users complain about it quite often: here are some references. Call it what you will, but a lot of people seem to disagree with per-repository pricing for small low-usage repositories, just because they are private.

If Dropbox Used GitHub’s Pricing Plan -- http://news.ycombinator.com/item?id=2780955

Bitbucket now rocks Git -- http://news.ycombinator.com/item?id=3066828

"Yeah, it's ridiculous. I have enough repos that I'd need the Platinum 200 dollar a month plan." -- http://news.ycombinator.com/item?id=3023826

"For Git projects, the per-repo pricing of Github is insane. They don't even have a public pricing plan for the number of repos I'd need for my personal (and private) coding." -- http://news.ycombinator.com/item?id=2740757


It's brutal when you think that if you made one giant repo they'd do it for almost nothing, but because you picked this other layout it could cost you $20-$50/month.

I want prices to be based on resource usage for the most part. If I've got a huge repo and I'm always using it, that would cost more. If I've got fifty tiny and essentially static projects then they'd be almost free to host. Give me non-arbitrary pricing and I'll be there.


I (have a private Github account. Let me echo something, "... when you think that if you made one giant repo they'd do it for almost nothing".

In other words, if my project was "open" Github would cost nothing. That's burtal.

Frankly I don't mind paying for repo services. I used Assembla for years. But then I moved away from svn. My problem is the price compared to all the free accounts. I (and all the other paying customers) are subsiding the open repos.

Now that's ok. I do gain from that, since I have several repos (projects) I follow and use. So it pays for itself in the long run. Sort of like public broadcasting. It's the same business model; I donate money every year to both PBS and NPR for the pleasure of tuning in. And I know the guy sitting in the car next to me blasting "Car Talk" probably didn't pay a dime.

But I"m ok with that. Really. (sic) :)


That's great (seriously: it is); however, I think it is then important to realize (for others) that GitHub's business model is closer to that of PBS than CBS: they are, in the long run, relying on the fact that they are doing the world a public good, and when competitors with enticing "fair" price points enter this space more powerfully, it will be interesting to see if they manage to "keep up the donation drives" enough to continue subsidizing the world's open source. (edit:) ...especially, to continue the additional point I have been bringing up in this thread, when this competitors are able to "outrun" GitHub, because they aren't carrying around that baggage: they can focus, intently, right where the profit is, allowing them to rip as much of it away from GitHub as they can.


If anyone can come up with a law that basically forces companies who give up on free products to open source the code they no longer use, that would lead to amazing opportunities!


It's not just the code --which might as well be useless.

You also need serious $$$ in infrastructure to run it for the web masses.

This is getting to be a problem for FOSS. I mean, once you could get some universities, volunteers, etc, donate some hosting for repository mirrors and stuff, and it was enough.

Nowadays, a service that caters to the web crowds needs serious many to run with some capacity (and redundancy). So, say, GNU cannot just start a FOSS Gmail, or a FOSS Github service running on free servers and controlled by the community. Also see: Diaspora.

So, FOSS is mostly confined to the desktop or as server backend infrastructure, but cannot compete as SaaS.


Reddit?


Wikipedia.


FWIW, Wikipedia has serious issues with backend scalability: one of the reasons why their history UI is as weird as it is, for example, is that they really really really don't want users using it, due to CPU issues in the diff algorithm (which they now have written in optimized C, and it still causes them problems).

Meanwhile, of course, they are constantly running charity drives to try to get enough money to keep the lights on. I am not certain you'd be able to do the same thing if you were running an open email service instead of managing the worlds seemingly-premier encyclopedia.


Those are massively read (as opposed to write) oriented, the content is the same for everybody, and as such cachable.

Now, think something like Gmail, Basecamp, Google Docs, Facebook where that is not the case, and we're talking about an order of magnitude if not more expensive and powerful infrastructure needed.


Your timeline here seems a bit off. Google Code Search didn't launch until mid-October 2006.


The specific year is not an important feature of this story: whether it is five years lost or six years lost is irrelevant. I mean, the story isn't even really about me: it was about the entire ecosystem, as I personally am one of those people who believe "ideas have their time", and therefore that, in that year, a ton of smaller companies were looking at how to approach this specific problem.

That said, I was specific to note a gap between 2005 (when my friends at Okori, Brian Fox in specific: notable as he was the first employee of the FSF, developing bash, readline, gdbserver, and texinfo), and "the time we got organized enough to actually do it" (when Okori actually existed as a company that could allocate resources to building something like that); I therefore feel it is even more unreasonable to be bothering about something I don't remember a specific date of anyway (it isn't like I marked it in my calendar).


Google is like a big oil company that can afford to build a gas station right beside the mom & pop station, and undercut them until they go out of business.


And then they shut down their gas station, leaving the community stranded.


I'm very disappointed to see google code search go.

It was very useful (for me at least, don't read this as a comment on the whole committee / process) in finishing the new C++ standard, and answering the question "Well, did anyone ever really write code like X?" (the answer was usually yes).

Buzz makes a lot of sense, although I imagine some users will be disappointed it couldn't be more 'cleanly' imported into google+.


Same here on google code. I have used it on multiple occasions to find out how users are using a class or API to get a feel for what API was actually used for. Not to mention when learning a new API having the ability to see real world usage was always helpful. Not to mention the ease of finding apps that had security issues in codebases after it was found. Anyone know of a replacement for this?


Well github's code search is a pretty good replacement, although it's limited to github.


GitHub's code search lacks regular expressions, which makes it far less useful than Google Code Search.


I also found the code search very useful, and usually used it to answer a similar question: does anybody use a Perl feature that is about to change (due to a bug fix, for example) or for which changes were proposed.

I didn't use it all that often, but when I did, I was very grateful for having it.


Did they give a reason for shutting down code search? I didn't see it but think it deserves some kind of explanation of the decision.


It costs money. The kind of people who use it probably have adblock installed. It's not surprising that they would cancel it, they are a company driven by profit after all. I don't think there is really much to explain. Maybe once it had utility as a way to earn good will among developers, but presumably they don't care too much about that these days, given how little competition and how much brand recognition they have. (Is Steve Yegge is still wondering about that 'arrogance' thing?).

Every time you perform a normal Google search be thankful for all the people who browse the internet without adblock: they are subsidising you.


> It costs money. The kind of people who use it probably have adblock installed. It's not surprising that they would cancel it, they are a company driven by profit after all. I don't think there is really much to explain. Maybe once it had utility as a way to earn good will among developers, but presumably they don't care too much about that these days, given how little competition and how much brand recognition they have. (Is Steve Yegge is still wondering about that 'arrogance' thing?).

It does cost money, but I kind of doubt this is the reason.

First thing, I assume that Google will continue to have Code Search or something very much like it available internally. The reason is simple: Like many people here said, this kind of search is useful for us programmers. So Google would be silly to not let its people continue to benefit from it. Google is known to have lots of internal tools written to make its developers more productive - this would be just another one.

Second, given that assumption, I wonder how much money it takes to run Code Search publicly. The code is already working, development costs are already paid. How much traffic does it actually receive? I doubt it is that much that it is significant in any way for Google. Google's server infrastructure is very efficient, and the amount of people searching for code isn't that big in absolute terms.

So why would Google be shutting it down? I have no idea. I can throw out a few wild guesses, but don't take them too seriously:

1. Part of a push to limit the amount of Google websites and services - just to keep the number smaller, more coherent, more cohesive.

2. Keeping Code Search an internal tool gives Google another advantage that its developers have that others do not. 2.1. Perhaps Google just finished a major upgrade to Code Search, and decided it wants to keep that internal, and doesn't want to maintain two codebases.

3. Fear of potential legal issues, either 3.1. People suing on the assumption that code appearing in Code Search has been 'scanned' by Google, so Google can't say it never saw that code before, say, in a copyright infringement lawsuit (like the Oracle-Google lawsuit), or 3.2. People using Code Search to find similar pieces of code and using that to sue people. (Both of these legal fears seem ridiculous to me, but lawyers tell their clients to do lots of things that seem ridiculous in order to limit legal liability.)

Again, though, I really have no idea. But I doubt it is money.


It might not be simply a matter of money, there are other limited resources. For example there could be trouble staffing the project due to nobody wanting to work on it. Or be unwilling to staff it since they feel that anyone working on it could make a bigger impact elsewhere.

(And before anyone starts nitpicking that either of those issues could be solved by just throwing more money at it, money and people aren't actually infinitely fungible).


Interesting note: As far as I know Code Search was built mostly by Russ Cox, he is now one of the main developers of the Go language.


>>It costs money.

Then slap a Gogle Payments icon on the page and charge a dime per click. Or a nickel. They can make a pretty penny, no pun intended :)


As someone who routinely uses Google with AdBlock installed, I second this. For that matter, I would gladly pay Google a monthly fee to have unlimited search, if the "free" Google only had limited search. The fact that Google has chosen to make their money from advertisers instead of users like me is not my fault.


Yes its not your fault, but it's also not Google's fault. Your sense of entitlement is very surprising.


It's not "entitlement" to ask to pay for something.


Too bad Google can't do the opposite of acquiring a company for things like this. Google Code was incredibly useful, it'll be sad to see it go.


Corporate spin-offs are very possible and are done frequently.

The problem here is that Google Code Search wasn't independently profitable, so it wouldn't make sense to sustain it as an independent company. (I presume it was created to build goodwill for Google as a whole among coders.) As to whether any of the related assets are valuable to, say, another company who thinks they can make code search profitable, I don't know. But if Google can sell those assets to such a company, it will.


I'd guess that the main issue is that it's probably heavily coupled to the rest of Google's services. Their index, web crawler and general infrastructure would be vital to this product. If Google were to spin it off it would be useless without all these things.


Same here with code search. Why write code that already exists? Code Search helped me find exactly what I needed for a few open-source projects I worked on.

Google Buzz was dead before it launched.


I may be biased, but: I feel like Google's "geek cred" is slipping. It feels like PMs (and their "monetization strategies") are gaining control over at El Goog, shutting down anything that isn't "revenue positive".

You can't measure geek cred. You can't measure the second-order effect of services like Code Search.

So the slow slide of Google turning into "just another tech company" starts...


"You can't measure geek cred. You can't measure the second-order effect of services like Code Search."

And so as a publicly owned, regulated, billion dollar corporation, it no longer make sense to commit unmeasured resources to support unmeasurable benefits.


The funny thing is that the supposedly mature "we're a public, billion dollar grown up corporation now" approach almost always ends up robbing the company of its ability to lead and innovate.

Google is neglecting products it's great at (eg GAE) to focus on areas it's completely clueless about (like Google+). There's no real risk here, the ads are going to continue pumping cash in, but it's slowly becoming a big dumb company.

My bet is that within 2 years they will buy someone like Twitter or FourSquare for n*$100m, and you'll know the new AOL has arrived.


I probably agree with your first paragraph. That might be the inevitable consequence of big, or it might be too tight a focus on explicit shareholder value in the regulations, or ...

Your 2nd and 3rd paragraphs won't surprise me.


Every day companies like Google and Yahoo do things which don't have a directly measurable benefit.

Why does Yahoo put out YUI, Hadoop, YTS for free? Why does Google give out protobufs, GWT for free? Why are there public tech talks at these companies? Can you quantify the benefit of such activities?


What do you mean unmeasured resources?

Look, I admit that you can't make good business decisions on Google's scale without using the data. That, however, is not Google's problem. It's also important to have vision and principles that everyone in the company understands, otherwise it's impossible to keep thousands of employees moving in the same direction. It would be very easy for Larry to make a decision like:

"One of our core values is to put engineering first, this makes us a place where the smartest engineers want to work, and keeping Google Code around is a strong external indicator of this, therefore we should spend $1,000,000/year maintaining it regardless of profits."

Despite the lack of data you could not fault a CEO for making a gut call like that. Just because you can't measure something doesn't mean it doesn't exist.


"What do you mean unmeasured resources?"

Ah, sorry, I had saurik's comment in mind when I wrote that.

"2) I don't think Google even tries to make things sustainable. During all of these "Google shuts down X" that have been happened recently, some posts here by ex-Google people indicated that Google internally didn't even have reporting on per-project costs... hopefully now that they see how much certain things cost they will cause less market-level problems going forward."


Well, by that rationale, every company should just zero out their advertising budgets.

Seriously though, I see what you're saying, but there isn't always a direct metric for everything. No one fully understands a variety of drug interactions within the body, but does that mean everyone should stop taking their pills? Likewise, when technology companies stop caring about the community that develops software for their product(s), although it doesn't immediately affect their bottom line, one can usually point to it as a contributing cause for their slow decline (think IBM, Microsoft, Oracle).


And so as a publicly owned, regulated, billion dollar corporation, it no longer make sense to commit unmeasured resources to support unmeasurable benefits.

What's the point of making a billion dollars if it doesn't let you work on interesting things that don't offer any obvious immediate payoffs?


I believe as a public company, the only point of making a billion dollars is shareholder value, by law. Before they went public they could indulge their geek kink until they went blind.

I'm not saying Code Search doesn't (necessarily) contribute to shareholder value. But it sure has to be easier to make the case for something that can be measured, as opposed to "geek goodwill."


I also get the impression that Google's glory days may be drawing to a close. Certainly my level of trust in their services is not what it once was.


Google Code Hosting is another source of geek cred. They don't monetize it. No ads and no paid hosting.

I hope the lack of monetization is not their only reason for killing Code Search.


"I may be biased, but: I feel like Google's "geek cred" is slipping."

Thank you, Captain Obvious!


Sad to see Code Search go. I just tried a simple search for 'pthread_t' on both Koders (http://www.koders.com) and Krugle (http://opensearch.krugle.org). Both are mentioned elsewhere in the comments as possible alternatives.

Krugle found no results. Koders found results, but the response time was very large.

Both have a long way to go in being a viable Code Search replacement.

Hopefully Code Search just gets rolled into the primary Google search product.

NOTE:

I was just looking over the Koders results. It is tokenizing 'pthread_t' as 'pthread' 't', so the top results are not what I would consider useful. I'm sure I can change some settings to get proper tokenization for my languages identifiers, but that is more work up front.


Double quotes seem to stop it from tokenizing, e.g. "pthread_t".


Yes, but in a code search engine it would make sense to not break words that are valid symbols in more programming languages than not. It's all about sensible defaults.


There's a lot of significance in function/method signatures and non-alphanum chars that google is getting to grips with (and hoogle/hayoo for haskell seems to be state of the art).

Ruby Class#method, the (dozens or more) symbolic "operators" in scala, haskell, ocaml (which are generally functions, not operators).

I think it's up to each language community to build those tools now.


There's a competitor to google code search at least: http://www.koders.com/

Actually, I'm kinda glad they're phasing stuff out. It shows courage to do that, a lesson probably hard learned at google. They have a myriad of products, but would probably do much better with them if they thoroughly finished them before release.


Yes, one of the things Larry Page talked about in the earnings call yesterday was aggressive project pruning and focus. I agree that this is a good sign of Page's leadership.


The danger lies in being a company with a reputation of launching stuff and then not sticking with them. At some point that starts harming adoption levels of new products.

Edit to comment below:

"Trying and failing is a far better outcome than never having tried at all" is a great mantra for individuals or even small startups, but I am not sure how it applies here. I am still bitter over Google abandoning Notebook (which I was quite a heavy user of) and canning Calendar Public Search (search!!). I then reluctantly allowed my team to use Wave, only to have that one canned after building a process around it. No, it is not better for me that they tried and changed their minds (they did NOT "fail" technically), because I could have picked another product at the time that did intend to stick around.

People look at Google, with all the money and people that it has, and ask "How much could it POSSIBLY cost to keep Code Search going? How is it possible that Google can't afford it?". The only answer I can come up with is (1) It can afford it (2) But it doesn't care to.

So sure, if you are happy to see Google launch stuff that it has no commitment to, great for you. I'm having better luck taking chances with startups and their products. At least they have skin in the game.

Edit #2: @joebadmo - but that is exactly what I think when I see a new Google product - "will it reach such large levels of adoption so that Google will keep it". And I keep getting it wrong - even for products like Wave and Code Search which I thought surely they are used internally and so will be kept around if only because of that - nope, either they are not used internally (if not, why not?) or even that is not protection enough.

Sure, startups and small-person concerns are also risky propositions, but let's not forget the numbers. 100,000 users are a "failure" for google, but they are a wild success for a small startup, at least the ones not looking for a big exit. If you inlcude app writers as "startups", their longevity is phenomenal - I have never had an app stop working on me because somebody turned a sever off somewhere. Perhaps I have been freakishly lucky. The only other product I have had to stop using because it was pulled was wesabe.com.


I would definitely not base any sort of important business function on any free Google beta product without seriously considering the difficulty of migrating upon shutdown of said service.

I assume that a major criterion for Google shutting down a service is low usage. For them, that constitutes failure. Very few people were using Notebook, Calendar Search, Wave, or Code Search. I'm sure it's fairly simple to calculate the cost benefit of devoting even minimal resources to a product that 1. doesn't make money; and 2. doesn't substantially make the web better for some minimum threshold of people.

Google is one of the richest companies in the world and could afford a lot of things, but that's not justification that it should.

You're right that there's an interesting tension between starting/failing fast and getting people to trust that a product will stick around. But I don't actually think that's a big part of the calculation for most people. When Google releases a new product, I usually go either "Ooh, that sounds awesome, I'm signing up for the beta immediately," or "Not interested, moving on." I rarely go "I'm not going to use this because it probably won't get enough users to last."

I'm glad to hear that going with startup technologies is going well for you, but the danger seems just as big. Startups fail all the time, and you have no recourse afterward.


> I assume that a major criterion for Google shutting down a service is low usage. For them, that constitutes failure.

I guess the question is, low usage compared to what? Compared to other Google services (search, mail), almost anything will be very low usage.


I would guess the other half of this initiative would be to only launch projects that have plans that involve profit. Like I understand killing something off if it doesn't catch on, but I don't see how Google Code Search could ever make Google much money over their normal search.


I hope you're wrong, because I kind of like how willing Google is to start crazy projects.

I also think that making money directly shouldn't necessarily be the sole motivator for projects for Google. Google makes money as people use the web. The better they make the web, the better it is for Google.


I second Joe's sentiment here. I would much prefer discussing how Google abandoned a project, rather than never having tried it at all. Trying and failing is a far better outcome than never having tried at all.


I agree with you to the extend: some products needs to be cut.

However, the problem is that Google is synonym for search and getting out search business even for niche is a very dangerous move. That allows small companies to attack them from low-end and then slowly growing (from code-search, they add bug search, then add stack-overflow search, then social search, ...) and eventually they can become a Google replacement.


Very saddened to see 'University Research Program for Google Search' go. Google's index offers an exciting corpus for linguistics and AI study.

The current search API's just don't cut it for proper research (for example: just 64 results per query and 1000 queries a day [1] and the "estimatedResultCount" being off by a factor of 10-100 [2]).

I believe spammers were abusing the Google translate API to spin articles in different languages. This contributed to it being closed down. I don't hope that Google's search API is crippled to thwart the bad apples. Because then those that follow the TOS (don't crawl Google's results) have little recourse, but to halt their research (Yahoo Boss and Bing Api give little solace).

[1] Too few for either deep analysis or learning queries like:

  "X is a *" and "X, such as *,"
[2] Estimated results for "test". With API: 257.000.000 vs. manual search 2.750.000.000



Off topic but in response to your conjecture about the Google Translate API:

My own is that they are protecting against the harvesting of their AI algorithm's output.


Google Code Search has been pretty broken (for my use case, anyway) for a while. I build Drupal Code Search[1] on top of their API back in 2008, and a few months later, they stopped indexing code from Drupal.org. Since then, 2 new versions of Drupal (6 and 7) have been released, and none of the new code has been indexed, making my site largely useless except for legacy code searches.

I guess I will just shut it down completely come 2012, I don't have any way to do grep-style searches at the same speed Google's API could.

[1] http://drupalcodesearch.com/


Was it broken in some way other than it didn't index drupal.org?


Drupal.org most likely was not the only site they no longer indexed. Probably no code site was still being indexed, at all.


No! I loved code search, I'd use it as evidence when proposing the deprecation of API methods in open source projects :)


This is awful. Code Search is an indispensable tool for finding reference code and real-world uses of various libraries. I don't know of anything on par with it.


GitHub search is pretty good - it only searches code that's been hosted (or mirrored) on GitHub, but that's a LOT of code.


IMHO Github code search totally sucks. It's better than nothing, but it feels like the bare minimum - just a normal fulltext index applied to source code.

- It's not case-sensitive, which sucks for searching code.

- It's unable to search for any punctuation characters (searching for "foo()" is the same as searching for "foo"), which totally sucks for searching source code.

- The search finds lots and lots of duplicates. Even multiple matches in one file are listed as completely separate results.

- You can limit the search to one of the languages in dropdown menu, but if your language of choice doesn't happen to be in there, you're out of luck. For example you can't limit your search only to C header files.

Simply by adding package:github.com to Google Code Search search box one can apply a better tool for code in github, but unfortunately the good times seem to be over soon.

Really sad to see it go...


I'm not familiar enough with either to know if they are of equal power, but an alternative to Google Code Search might be Krugle open-source search:

http://opensearch.krugle.org


I wouldn't miss Code search much if the regular Google search would understand regular expressions or terms with underscores. But it does not, and also increasingly annoys me by misinterpreting my search terms and searching for the things it believes I meant instead of the things I told it to search for.

And while DuckDuckGo has a much nicer search frontend, it's Bing-fed index sadly sucks, making it no universal replacement for Google's declining frontend.


Headline should be: Search company shuts down search product to focus on social networking.

I suppose their Blog Search is next.


I'm actually really surprised that code search is getting phased off. It's a great resource to see how libraries are implemented in the wild -- it also exposes some common errors that lots of library authors make.

So sad to see this one go -- but I think it will make for another opportunity to allow a competing site like koders.com to iterate on building a product that developers would love to use...I hope.


It's too bad about code search - that was really helpful for finding example code (and linking to specific snippets of code within open-source projects!)


Hate to see code search go. I always found it useful when having problems with some less-well documented libraries to see how they were used in the real world.


I'm surprised about the Code Search news. That seems to be squarely within Google's mission.


How do you sell ads on a bunch of code results? That's the only logical explanation I see for shutting this down.

I find a globally-searchable database of code useful enough to want to rewrite this myself. I won't get all of the Internet, but if I can get Github and GNU / Apache / etc., it will be enough.


i"ve only thought about it for about 30 seconds but it would seem to be the absolute best place to post engineering job ads.


Also - ads for e-books perhaps. If you end up looking at Hibernate code, you are struggling with Hibernate. Why not clicking on "Hibernate In Action"?


interesting source code centric adwords can be imagined here - bidding on specific language, api, library, product :)


At Pattern Insight (http://patterninsight.com) we have build a source code search engine. We do not give this to our customers as a standalone tool most of the time but it's the underlying technology for our product, Code Assurance, which helps companies eliminate bugs from their releases.

We use it internally to search our code / libraries, if anyone is interested in indexing/searching his own code, especially if it's open source, I would be happy to provide a copy. Email: spiros at patterninsight.com


It never worked all that well for me.

What I want is a "curl | grep" for the web. Just something that searches the entire page, including <head> <!--> etc. I can do without fancy semantics.

But Google tries to be smart even on quoted queries. And that annoys me deeply.


>But Google tries to be smart even on quoted queries. And that annoys me deeply.

They're just trying to redirect you to more profitable (i.e., a greater probability that you will impress an ad) search queries.


Really?

"Google's mission is to organize the world's information"

How is so when code search is going away? Google did excellent job at indexing the code, so why to throw away what's already working?


Maybe they don't believe in the Lisp notion that code is data.


I can understand about Google Buzz ... they are replacing it with Google+. But what about Code Search? It's like they mention it once, and then don't give a reason. I think that was useful for many people! I wonder why they are closing it...


Speaking of focus, we went from

"Organizing Worlds Information"

==>

"Organizing People in to Circles"

Indeed exciting times!


Also: there will always be people decrying Google's tendency to launch and phase out new products and there will always be people decrying its reticence to try new things even if they fail, and they're probably doing a good job if they're attracting roughly equal amounts of these responses.


Buzz is fine. But what was wrong with Code Search?


buzz was awesome for link sharing via a bookmarklet to my professional audience (compare to facebook bookmarklet for life stuff). plus isn't there yet -- plus doesn't expose RSS feeds to work with my audience's existing workflows. damn.

any word on a plus RSS api? i want my stream, and i want my +1 feed. i was looking into the +1 stream yesterday, seems like google made it as hard as possible to hit from javascript -- nonstandard http post, no JSONP = i don't know if its possible to do client-side. damn x2.


https://developers.google.com/+/api/latest/activities/list

It returns JSON (and you need to OAuth it), but it gives the stream info you are after.

https://developers.google.com/+/api/latest/activities/search (with your name) is useful too.


Hey dustingetz,

You might like to take a look at Handpick:

http://handpick.me

You'll be able to create different groups of people to collect links for via a bookmarklet throughout the day and send them via a single email at the end of the day.


I used code search just yesterday. Super lame.


If Google can't backup the resources required for Code-Search, then who will ? Just googled to see if any other code search service is available. It listed one more result other than the Google Code Search. Will koders.com remain alive. It will probably be a single. Lets hope, there will be someone backing up Koders.


Yes, Koders is here to stay and will be relaunched with several improvements in near future. Meanwhile if you have suggestions/wishlist of improvements, you are welcome to leave a comment here: http://news.ycombinator.com/item?id=3132462


Does anyone know what they mean by this?

"we will remove iGoogle's social features on January 15, 2012. iGoogle itself, and non-social iGoogle applications, will stay as they are."

Are they referring to Google Chat or are they referring to iGoogle widgets that have "social features" (e.g. Twitter widgets, Facebook widgets, etc.)?


Some gadgets have social features so interact with friends who also use the same gadget. For example there could be a gadget that lets you search for and watch YouTube videos and friends you allow can see what videos you watched. Those gadgets are going away. You Chat will still be there and gadgets that interact with third-party social sites will still be there too.


i don't understand why codesearch has to go. granted, it felt a bit neglected lately, but it was truly useful.

i wish there was a way for google to open source their abandoned projects. i'm sure someone would be willing to offer a similar service by basing it what the google code search already does.


I hope Google open source the code behind Code Search. They don't have to release all supporting libraries that are specific to Google's infrastructure though. Such that those of us who actually use the service can figure out a scaled-down implementation to serve ourselves.


I liked Buzz's idea of integrating a social inbox with the an email inbox, and in particular I liked Reader integration with Buzz. If Google doesn't integrate Reader and Gmail with Plus more effectively, I'll be spending _less_ time being social with Google than before.


I could see a product similar to code search being integrated into services like github and others. Code search wasn't perfect, it basically just searched code. What if you had a product that could, like an IDE, follow method declarations and the like? That'd be cool.


Why I expected Buzz to be a failure, based on a cursory UX analysis & comparison with existing social products: http://alanhogan.com/buzz-is-already-dead (Feb. 2010)



What is the best alternative to Google Code Search?


Recently I have found out that google search "function/class/variable/anything else filetype:extension" works even better than google code search.


I don't know how well does code search work for searching punctuation but I'm sure general web search doesn't cut it.


Code Search allows using regexps for searching, which is one of the things that made it super awesome.



Thanks. It seems like these two options are way behind Google Search. Maybe there is opportunity here to build a pay service for developers (like to find implementation patterns, common bugs, etc.).

Anyway, I just want to express my disbelief with this google decision: Google is synonym for search. All kind of searches.

Meaning if Google cannot provide good search for open source code then something is terribly wrong. When can say everything about Google and some failures of Google, but they are very very good in search. And, they want to be around for next 15 years, they must not lose on that front (even if it is just "open source search").

Anyway, I hope DuckDuckGo or some other search engine will try to do something here...


I'll often just do a GitHub search. If you're looking for something mainstreamish, there's probably some kind of public code using it on GitHub


Nullege is still great search engine for python code.

http://nullege.com/


Louis CK (you might know him from the popular "Everything's Amazing and Nobody's Happy" video) frequently appears on Opie & Anthony. He makes a similar point regarding chains driving out local stores then closing up shop in this show: http://www.youtube.com/watch?v=N95IMKRkcBw


Great. Now what are we supposed to use to search code? I can't think of any alternatives.


does this mean google reader sharing now goes to plus? Me and a close group of friends use google reader a lot for sharing and commenting, personally more than i ever use plus or facebook.

sharing on google reader also shared on google buzz as well though.


My friend group is the same, but this was always annoying because Buzz didn't go back into Greader. So I had one friend who would post to Buzz instead of Reader, but I would miss his posts unless I actually went to the Buzz interface, where I would still have to wade through all the Reader shared stuff.

I'm glad it's all moving to +. I can only assume that sharing from Reader will be connected to + at some point. There's currently a workaround for this, but it's ugly.


i also liked buzz for bringing reader shares / comments into my inbox / buzz feed...i think the value of buzz was underestimated. i hope google brings the same integrations to google+. external sites were the bomb!


Anyone knows the reason?

I thought that it's their robot working hard (no human resources allotted for it)


That's what I really like about Google. A company with too many products just hanging in there making losses is a company with a low morale. I think accepting the loss and shutting down low performance products is very important.


That makes me sad. I use it a lot!


Damn. I'll miss Code Search.


Oh no, not the code search!


Google's profit margins greatly exceeded expectations for the last quarter[1]. In response they are shutting down Code Search?

[1]http://www.nytimes.com/2011/10/14/technology/google-reports-...


boom




Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: