194 comments

[ 42.4 ms ] story [ 3966 ms ] thread
update:

after hour currently at 675.01. Price increase +73.01 (12.13%)

With a market cap of half of Apple's they made ... 1/20 of the profit. For the first time ever.
what are the growth prospects for Apple vs. Amazon? Market cap is influenced by speculators who want to buy low and sell high, and to them - the bigger the growth prospects, the better.
AWS currently improves product and cuts prices. My feeling is that as they mature over time and get even more of a lock on the market (and their customers) with the obvious lock in they have that will reverse and prices will increase over time resulting in even more profit than they reported today. Impossible to think this won't happen, especially with the differentiation they have with their product as well as the FUD of changing to a marginally less expensive provider (who will probably follow suit).
Or another future - as more people see Google and Azure can compete... they build cloud agnostic software that treats providers as a utility, and profit margins fall and fall.
> they build cloud agnostic software that treats providers as a utility

It's like browsers all over again. As long as each provider is offering different things or different implementations of the same thing we will either need to choose one or be forced to go through a lowest common denominator translation layer.

MSFT, GOOG, and AMZN are all smart enough to not turn this into a race to the bottom of who can provide the cheapest scalable instances. They'll all try to differentiate on features and so it's going to be a long time until you can treat these as a utility.

I think you're right in that Amazon Machine Learning and Azure Machine Learning will never be drop-in replacements for eachother.

However, I would argue that VPSs are already drop-in replacements and I bet (though I have no proof) that VPSs are the majority of the market. The cloud providers may not have a viable way to provide even the perception of vendor lock-in when it comes to VPSs.

Now it could be that prices and profits stay high anyways as the market is an oligopoly. The U.S. has experienced the same thing with mobile operators. This is also far from obvious though because entrance barriers for VPSs don't seem steep.

Sure, a basic VPS could be a drop-in replacement, but think of all the stuff that goes on with managing that. If you want to automate the creation of new servers then you have to script it or use something like CloudFormation or Troposphere. Either way you are immediately off the "drop-in replacement" path unless you build a middleware layer for your scripts.

These guys make their money off the companies that need to be able to automate infrastructure and that stuff is becoming more, not less, fragmented.

Even bigger than VPS is storage, which is the closest thing you will get to a commodity/drop in replacement in the cloud.

It was recently reported that Apple was spending hundreds of millions of dollars on S3 (before they switched). I think it's safe to assume that a couple of billion dollars of AWS revenue is cheap commodity storage.... which has low switching costs.

Well so far MSFT and GOOG are regularly slashing prices to keep up with AMZN's price cuts. So we're in the price war stage still, not much at the differentiation stage.
> So we're in the price war stage still, not much at the differentiation stage.

Depends what you're looking for I suppose. If all you need is basic compute instances then, yes, it's a price war. Any products beyond that have a tenuous relationship between the different providers.

Even something as simple as using a columnar DB pretty much locks you into one of the providers because you have to build around all the nuances of SQL Server vs. RedShift (Postgres) vs. BigQuery.

Forget talking about things like infrastructure automation once you're trying to manage this on demand across many environments in a reproducible way.

This isn't correct.

Amazon has not matched our May 2015 price cut for GCE (they instead have tossed a couple 5% cuts). They responded to our 2014 cut a day later. We saw a similar reaction with Nearline last year; a few months later they introduced S3 Infrequent Access, but it has plenty of caveats (rounding up to 128 KB per object) and still a 25% higher price per byte.

In the meantime we introduced custom shapes for GCE giving our customers a measured 20% savings. Finally, even Microsoft has per-minute billing.

If you're looking for the best price/performance you'll have to come to Google Cloud.

Disclosure: I work on Google Cloud, so of course I want your business ;)

You have it backwards. It's Google that's slashing prices. The others are begrudgingly reacting.
One issue with this is the idea of competition from physical hardware.

At some price point, if the prices are too high, it will be cheaper to hire extra people, and get a cage in a few datacenters and some physical hardware.

The problem is people who arent big enough to have a whole data center. Or dont have the upfront capital even if its benefital in the longterm. They will still rely on Amazon.
You don't need to own a whole datacenter to own machines located in a datacenter, my company keeps machines in a 4 rack cage in a datacenter. (and plans to empty that cage by the time our contract runs out in September to finish the migration to AWS). Our AWS bill is just a bit higher than what we were paying for the cage rental + internet connectivity alone.
That doesn't sound possible. Are you saying your raw bill for _just_ colo (ie. space and power) and bandwidth (ie. IP transit) was less than your _entire_ AWS bill? I've never heard of anyone take such an extreme stance.
The 4 rack cage plus extra power plus redundant network at a Tier I datacenter costs us close to $6K/month (including some block of remote hands hours). Even at our peak we didn't have the racks full, allowed room for expansion. Note that this cost includes only coloc costs, not server or network leases/purchases.

We have very bursty traffic and run our baseload of ~ 30 severs (mostly m4.larges, with a few smaller utility servers as well as larger servers for back end services) most of the time, but when we need to scale, we ramp up to about 2 - 4 times that, but usually only for a few hours at a time. We also do some batch processing that takes a few high-memory servers for about a day each week.

We monitor costs pretty closely and our monthly AWS spend is still less than the coloc fees, and we are much more scalable now and have multi-datacenter (AZ) redundancy, and by the end of the year, we will have a cold spare in a second region (which will cost us very little since it will be mostly powered off)

We no longer need to vastly over-provision servers to handle our peak load and batch processing needs.

One significant advantage of AWS is that they have a very large, built-in retail customer with relatively demanding requirements. I've always been under the impression that Microsoft doesn't really have a business that would lend itself to consuming its own services[0] and it's unclear to me whether or not Google uses their stuff internally.

0 - as an anecdote, I recall their salesmen coming to sell my employer their caching solution around 2008. I asked the rep I'd Microsoft was using their caching product on any of Microsofts web properties and he replied that they were not.

Xbox live runs on azure
Certain, newer pieces of some of these services, such as the Azure Active Directory service at the heart of Office 365 and some of the newer, complementary Xbox Live services are hosted on Azure. But the core Bing, Office 365 Commercial and Consumer offerings and Xbox Live services are not hosted on Azure. (same article as above)
It seems like amazon is using an "all in" strategy which seems to be working out pretty well. They first focused on porting the entire amazon retail website onto AWS which I believe they completed in 2011. They also went "all in" with the service oriented architecture for AWS. Both of these have paid off quite handsomely.

Granted, that hasn't always paid off (we all know how the fire phone turned out), but the wins have definitely made up for the losses.

Office 365 relies on Azure as well (unclear how much of it, though)
A number of Microsoft's own services -- including Office 365 business, Office 365 consumer, Bing and Xbox Live -- are not running on its Azure cloud backbone. They are still running on their own custom stacks and datacenters.http://www.zdnet.com/article/microsoft-is-on-a-quest-to-move...
Sentence immediately after the two you quoted:

Certain, newer pieces of some of these services, such as the Azure Active Directory service at the heart of Office 365 and some of the newer, complementary Xbox Live services are hosted on Azure.

Go cloud go!!
On one hand I miss my RSUs

on the other hand I do not miss working there.

well maybe a little. it was fun for a while.

Same...But good to know that it is executing well.

Go google cloud go! Bring AWS price down!! We need competition!!

Google indeed seems the strongest competitor now for AWS, but there are far more providers who compete - see a comparison: https://www.cloudorado.com/ . We just need to choose them to avoid monopoly or at least oligopoly.
Google stronger than Azure?
I think momentum wise yes...

Google is more a service company than MS, and I believe it is more experienced in cloud technologies.

I don't know, my experience with Google service is that it is on a par with Microsoft UI - makes me want to break things. Maybe we have different meanings of service though - I am thinking of service as 'something does not work, can I get some service for this thing of yours I am using?'
Google Cloud is amazing. Azure is terrible. Azure js for companies that, 15 years ago, were spending their budget on Windows. Great for "enterprises" that want to "leverage cloud".

Azure is slow, expensive, poorly designed (network and SSD designs are just bad), and the real slap to the face is that the management UI Is worse than anything from Windows 8.

"enterprises" that want to "leverage cloud" is where all of the money is. AWS is chasing that dragon too. There is way more money selling cloud stuff to Coca-Cola and GE than to software companies.
Yeah I'm sure this can work out well for Microsoft. But let's not pretend they're in any way a good fit for the scenarios gcloud covers. I'd wager for most of the people on HN, or at least the "startup" focused ones that want a solid IaaS, Google blows Azure out of the water.
True. But selling to those companies, can hardly cultivate an ecosystem around it.

Without vocal early adopters and evangelists in the wild, it is impossible for AWS to get to its current stage. Those people may not bring in that big of money, but they are willing to take risks and eagerly to give feedbacks, sharing their experience and come up with their own solution around it. The feedback loop is enclosed in that way, and the service can evolve with it.

MS still doesn't hit that spot yet.

I wouldn't touch any Google product that my business critically depended on with a 10 foot pole. Their non existent customer service, willingness to kill products if it does not meet their user or revenue expectations, among many other things. It's a good strategy for them, but bad for their consumers.
I think your comment is fair but maybe misplaced in this case. Hosting advanced services isn't a new market. Amazon has demonstrated there is demand.

All Google needs to do is innovate on tech, pricing, and marketing. They can chip away at Amazon's lead. Google does not need all of Amazon's customers, just some. And, it's a growth industry.

My experience with GCE customer service has been < 10 minute replies.
Strange to see 8 upvotes 16 hours after I posted this... are some googlers out there on an upvote spree? :)
I hung onto mine... Thinking this might be the catalyst to sell the few that I have.

It's a shame they have such a toxic environment and backload their stock grants. It would have been nice to be able to get rich and retire young!

I worked there from 2004-2007.

I'm glad I don't really remember how many RSUs I sold at $40.

May 15th is going to be a nice day for a lot of people ;-)
(comment deleted)
What is the significance of May 15th?
Employee stock vesting day. If you get stocks granted as a bonus (most of us do), they vest in may and November.

I know guys who've never sold any of their stocks. This will be literally $10k or more in their pockets, if they finally decide to sell.

Jeff just made $6.1B afterhours...
It's "papers" money. It's not cash until cashed out.
If he's giving any of his "paper money" away, I'm first in line.
The vesting schedule at AMZN is 5% after 12 months, 15% in 24 months, and 20% for every 6 months thereafter. For me (SDE I), that works out to $2200 in year 1 and a projected $7000ish in year 2. MSFT, on the other hand, has their stock vests 25% after 12 months and 25% for every 12 thereafter.
(comment deleted)
> If you get stocks granted as a bonus (most of us do),

Or by bonus do you mean just the amounts they give you as regular refreshes?

AMZN doesn't give cash bonuses; the RSU refreshes are treated like a bonus that pay-off in the future
I was mostly just curious if RSU refreshes was what was meant by bonuses.

Just never heard the term bonus used to describe them, but that makes sense.

Do other places typically do additional stock grants (bonuses) that don't vest for 2 years?

Does that event generally lower share price on the market?
Non-executive employee stock comp is a tiny fraction of total stock. Yes, the sales will increase supply and push the price down by a pitifully-small amount, and an institutional investor will buy them and push the price back up.
Probably when stock vests for college hires.
All hires
That is not 100% true. My AMZN stock vests in October for example, granted I'm an odd case.
Once you start getting part of your regular comp in stock, it will vest on November 15th, and May 15th. This may or may not occur until after your initial grant has vested depending on your yearly total comp (your manager should be telling you this ;-) ).
Is there any reason why Amazon would suddenly "waste" money on profit? Could this be an indication of a change in strategy or is this just a temporary thing?
Bezos will hold a press conference "We are sorry we did not realize we would make as much money as we did - consequently our spending just could not keep up" while AMZN tanks.
Bezos's goal seems only to be Amazon's growth and efficiency. I'm really starting to believe that being one of the 10 richest people on the planet was just a happy side-effect that only distracts him from his goal.
There should eventually be some time when Amazon does finally cash out, presumably on some timeline as a function of Bezos's mortality. Its unrealistic to be around forever. How many companies last hundreds of years? How many tech companies last decades?
> How many tech companies last decades?

Amazon is at 2 decades and going strong. Any reason to think they won't last a long time?

My prediction is AWS will split after Jeff dies (or is otherwise not involved in the company). Then Amazon retail will split from Amazon logistics, and so on.
Microsoft is 40 years old. IBM is 100. Some other old companies might not be considered "tech" anymore, but only because their market/product are better defined (e.g. Ford is an auto maker, but the assembly line and the concept of an automobile itself place Ford as the "tech" darling of 110 years ago, IMHO). I think if you don't see old tech companies today, it's because over time their public appearance has quietly lost the "how" (new technology, automation) and only left the "what" (search, ipods, advertising, rockets). But that doesn't mean old tech companies die.
Survivorship bias. The vast majority of companies die. The vast majority of tech companies die.
Allowing AWS to print money, resulted in adding nearly $150 billion to Amazon's market capitalization.

Investors have placed an immense value on AWS in the last year plus, since it was revealed how much operating income it was spitting off.

Bezos had been looking for a higher level cash machine (AWS has seven times the margins of retail) for a long, long time that he could utilize to step up to another level of financial capabilities and competition. Something that might sustain his ability to fight with Google and Apple. Fire Phone was meant to be that for example, which is why they attempted to charge premium prices; they thought they could slice off some iPhone type profit out of it.

It's interesting to note that Amazon third-party seller unit share is at 48% of total units shipped, which is a record high. It has grown by 1% per quarter for the last 4 quarters in a row.
I don't really like how they sell things from third parties on there. If it's sold by someone else and fulfilled by Amazon that's ok, but otherwise the whole point of buying on Amazon is to not have to deal with weirdo possibly-shady little retailers.
Amazon is selling trust there; in my experience, if the seller has a 96% or better rating you'll seldom be disappointed when it arrives, and they'll work hard to give you satisfaction if you aren't.

Much better than eBay in my family's experience.

If you have Prime, and click on that as a filter, that gives you fulfilled by amazon results... A lot of times I'll go another step and choose Amazon.com as the seller.
Amazon is pushing very hard to increase Fulfillment by Amazon coverage across the catalog, and is also releasing new products such as Seller-Fulfilled Prime. The customer experience is tightly controlled in both cases.
honestly - stuffing video in with prime then raising the price. recently offering video stand alone but for $9 more a year than prime (with video)... Hard to tell who is actually shipping an item - sketchy pricing based on region / or account etc..

why trust them with aws? it seems their strategy is to price things stupid low - get you on board then raise it when they think they 'got you by the balls'

no thanks

Jump on board, but don't wear the shackles! Use something like Kubernetes and FOSS projects and you can shift to whomever has the cheapest VMs today.

Easier said than done, I know, but a fine goal to have, I'd say. :-)

I totally agree with you in the ideal case. Problem is in some ways, AWS is too awesome due to all the add-on services.

Systems are more than just compute, and the last 3 companies I've worked for have become increasingly coupled to AWS over time due to use of the add-on services.

You build your system and over time you need message queuing, ability to send emails, databases, load balancing...

You can run / build these yourself but do you want to use up your team's bandwidth setting up say, postgresql replication ... or just use RDS?

If you use third-party (e.g. out of AWS) services, you lose the ability to do security with say, instance roles and in larger orgs you lose time to the "new vendor due diligence" process and getting the business to sign off on new support contracts etc. Maybe you also need some new automation tooling.

Because all that business-ey stuff is already in place with AWS and administering say, event queues, is sure not core business, most project teams will SaaS that capability if they can. Whatever service is already in AWS ends up being the no-brainer answer.

So hard to avoid architectural coupling. Then you look at the devops scripts and realise all the usage accounting, security configuration, deployment scripts and so on would need to be rewritten, and you have to go through compliance again if, hypothetically, you were to move on from AWS...

Apprenda does all of that, without relying on a company far larger than you
> Starbucks To Begin Sinister 'Phase Two' Of Operation

> SEATTLE–After a decade of aggressive expansion throughout North America and abroad, Starbucks suddenly and unexpectedly closed its 2,870 worldwide locations Monday to prepare for what company insiders are calling "Phase Two" of the company's long-range plan....

> Though the coffee chain's specific plans are not known, existing Starbucks franchises across the nation have been locked down with titanium shutters across all windows. In each coffee shop's door hangs the familiar Starbucks logo, slightly altered to present the familiar mermaid figure as a cyclopean mermaid whose all-seeing eye forms the apex of a world-spanning pyramid.

http://www.theonion.com/article/starbucks-to-begin-sinister-...

Whoops, promoting your cynical humorous reference while this was trending might have caused a backlash, sorry.

For the record I still want to buy Amazon stock someday.

Amazon's rationale for its lack of profits is well known... Why does this rationale not apply to its cloud business?

Is the cloud business so good that it generates profits regardless of the investment it makes... or is the retail business so bad that it can't generate profits regardless of the investment AMZN makes in it?

The retail business is competing with a very old, very established industry. The cloud business pretty much created a new industry.
I think its also true that their online retail store created a new industry. They might not be the first online retail store, but they definitely spearheaded the online retail industry. I think the some of the reason why AWS is more profitable or has been profitable longer because.

- Amazon reinvested its revenue on its growth.

- AWS was a by product of Amazon retail store infrastructure (at least initially).

- AWS customer base are mostly businesses with big pocket and less likely to worry about spending money.

- Amazon retail store is built on the business of selling products cheaper.

- AWS profit margin tends to be higher than retail store.

Just thinking out loud.

"spearheaded the online retail industry"

"Online retail" is not an interesting category even if they did spearhead it. Amazon is nothing more than a catalog retailer, something Sears Roebuck and Montgomery Ward pioneered eons ago. Shipping goods to consumers out of a warehouse by common carrier is not a new industry.

Neither is leasing out compute resources.

The devil's in the details. Scale, convenience, ease of use...

How long has leasing out compute resources been an industry? 1-2 decades? I'm sure there were leases on timesharing systems in the 50s and 60s, but that's still just a half century, and hardly the same industry as AWS.
It's not the same industry as AWS in the same way as Amazon isn't really the same industry as catalog/mail order businesses of yore (around 1850s I believe that began).

You could get limited amounts of compute time on a mainframe back in the day, just like you could order flannel and get it delivered in days/weeks.

Now you get massive scale compute resources on demand and fresh fruit, veg and almost anything else delivered within an hour.

Neither of them are dissimilar from present offerings, but neither are the same industry as their predecessors.

Considering Amazon is able to constantly expand into new retail fields (like groceries), launch a cloud services business, launch own its internal logistics business (that is the 2nd largest provider in the UK now) says to me the Retail business is probably more profitable than the media would have us believe...
From speaking with internal Amazon employees, their total profit is negative because they invest more money into expansion than they are taking in. They make an incredible amount of money and likely would be turning a profit if not for the extensive investments. But as it stands, expenses (including expansion costs) outstrip revenue, thus profit is negative.
Which is fine, so long as growth exceeds your debt-load. As it stands, it's one of two ways corporations should work. Either reinvest, or give out dividends... holding on to more money than your operational costs and/or shifting them to overseas subsidiaries to avoid taxes doesn't help the economy, or the communities the corporation services.

I'm okay with either disbursement or reinvestment/expansion/research. Those both grow communities/economies.

The impression I've gotten from friends working there is the reverse: The cloud business is making so much money it can't spend it fast enough. The retail side, meanwhile, is making money but spending it faster enough to remain "not profitable"
That's not really how accounting works
It's how it works at Amazon.

"How much gross profit is the retail business going to generate that I can spend on drones and the Amazon Echo v2"

For many years Amazon was proud that they did not make a profit.
This is an interesting question.

The e-commerce business seems to provide a near limitless opportunity for fixed investment towards getting your items to you faster. At what point will Amazon no longer have opportunities to invest in better logistics? 1 hour delivery for everything? There are hundreds of billions of dollars worth of investable opportunities there.

On the AWS side, I struggle to see a comparison. Yes, they need to build more infrastructure to scale the service, but there seems to be a saturation point. Where is the equivalent "last mile"?

Current global IT market is trillions of dollars, AWS is ~10B... lots of room there.
Of course there's room, but there aren't obvious ways to deploy billions of dollars. Products need to be developed for AWS, whereas for the e-commerce business there is a much stronger link between investment and ROI it seems.
The retail and video streaming are really just a massive distributed integration test for their cloud services. We're paying for the privilege of doing their QA for them.
It's the same exact thing they used Toys R Us and Target for, when launching their partner shops originally. With Amazon's extremely frugal nature and thin margins, they'll always look for ways they can get someone else to pay for their learning mistakes.
The switching cost for retail is really low; one Google and you know where you can get the same product for a potential lower price. This means that to win in the retail business you have to accept incredible low margins or even periodically a loss. Online retail is almost a 'perfect' market.

While with hosting, it isn't as easy to switch. When you have chosen AWS, of course you can switch to another vendor, but that is not without an investment in time and money. The higher the switching cost, the more you can raise the price above cost without customers (immediately) running away.

I would argue the exact opposite, but it's because it's not only just the product, but also the customer service and experience.

Switching from Amazon to a similar service is a huge risk to me. Will I get the same customer service when I order something and it comes shattered at my door? Can I rely on two day shipping being really 2 days? (I know some people here are whining about it being 3, the majority of us get it within 2, sometimes 1) Can I trust the reviews on the other sites? Will they know my preferences and suggest me other items or books as well as Amazon or are they coercing me? Do I get the same value added from the "premium" subscription such as Prime? Who knows.

Switching cloud providers? A snap, provided you have architected your systems according and not treated AWS offerings as black boxes and understand "OK I am using RDS, I can just switch to another MySQL provider easy enough" and prepared accordingly. "Ok I am using Elasticache, time for Redis." I imagine it's even easier if you stuck to containers. And unlike Amazon retail where there really isn't an online store that can match it in terms of customer service and experience, many would argue that a DigitalOcean or similar have better customer service, so there is competition there.

Amazon's cloud products apparently don't have to be at the low end of the price spectrum. And yet they continue to dominate. That's much harder to achieve in retail.

Another difference is that Amazon already has the dominant position in cloud computing. Amazon's share of all retail (not just e-commerce), is not dominant.

> Why does this rationale not apply to its cloud business

Radically different margins. AWS is running on a 23.5% operating income margin. Retail doesn't come close to that. Operating margins on their traditional North America business are 3.5%, and 1.3% internationally. Retail has to do almost seven times the sales of AWS to match the operating income.

The cash cow they have in AWS, is supporting the outsized valuation in their stock. Once investors realized what they had in AWS, by revealing how it was doing, the stock soared to a new level of high valuation. You can go back to December 2014 as see when this happened, the stock had been stagnant for a while, then doubled in one year following the reveal of how AWS was performing. AWS will pay for a lot of patience with shareholders.

I think you're right, that comparing operating margin is the right way to look at this. But do Amazon and its investors really believe that investing in video streaming, fire phones, etc will improve that margin?
(comment deleted)
Why do people link to basic stories like this behind a paywall? If WSJ has an original piece out, fine, but this could've come from any one of a thousand sources.

http://www.bloomberg.com/news/articles/2016-04-28/amazon-sal...

WSJ subscribers probably don't even think about the paywall when they share it here, since they are already logged in. But that is what the "web" link is for. If you see a known paywall site, always just click on the web link.
You can always google the title and get around the paywall.
No, that doesn't work anymore.

edit: the /amp trick mentioned below works beautifully, however.

(comment deleted)
Googling for the title works for me.
I just saw that too. Now I don't know what to believe.
It stopped working for a while but the re-implemented it recently.
WSJ subscribers may not be aware of this, but if you share an article via email, WSJ generates a shortened link to the same article that works for non-subscribers. Try this: http://on.wsj.com/1SNz3dT
That didn't work for me, I see the paywall.
It's probably limited to some number of click throughs...
The latest trick for wsj articles is to append /amp before /article

More readable, less ads, and no-paywall.

This was the only of the above "tricks" that worked for me without having to allow too much crazy javascript junk.

Side note: I hate any website that try to load more than 5 js scripts.

And it's even available in the page as a link: <link rel="amphtml" href="http://www.wsj.com/amp/articles/amazon-reports-surge-in-prof...

That's a call for a browser extension!

No need for a browser extension, just throw this into your ~/.js

    var ampLink = document.querySelector('link[rel="amphtml"]');
    if (ampLink) {
      var href = ampLink.getAttribute('href');
      if (href) {
        location.replace(href, false);
      }
    }
EDIT: woa! this works really wall ;)
This is something HN should just do anyway; that is high rep users flag a source as paywalled, and going forward the submitting user gets an alert with suggested alt links. No idea why HN isn't open source with an open source API to insert data too.
A variation of it is open source. The anti-spam features are guarded secrets.

They take a pretty conservative approach to adding features.

By variation, do you mean a clone. If so, that's not what I mean, I mean that the site is open source, community is able to make pull requests, and it's a community driven resource, that driven by content, code and community, not an singular interface.
By variation I mean HN is based on http://arclanguage.org/install ... I don't know how much the code bases have diverged.

They do have an API: http://blog.ycombinator.com/hacker-news-api

Yes, ARC is a great example of how a PG project should be open sourced and run by the community.

__

API is read-only; API should be "full service" and the first interface to HN, not a supplement.

___

Yeah, whole ARC (aka PG) thing is an issue too; FYI, those install direction are dated, here's the info: http://arclanguage.org/user?id=akkartik

PG doesn't even appear to be actively involved in ARC anymore: http://arclanguage.org/user?id=pg

Find it hard to believe if YC funded a startup like HN that they wouldn't require an explanation of why ARC made sense as the language of choice.

So....block ad blockers and HN screams.

"I will pay for content", they say. "I just don't like your 25MB flash ads."

Link to crap, and HN screams.

"Is journalism dead? This stuff is just click bait. I appreciate good writing and quality content."

Put up a paywall and HN screams.

"Why u make me pay?!?!?!?"

Solution: don't do shitty things (e.g. block ad-blockers, write awful content, or use paywalls). Easy.
To be fair, this doesn't address publishing's revenue problem, though it does the public good element. See my earlier comment.
Serve first-party ads that integrate well with your content, and establish direct relationships with your advertisers... same as print has done for over a century.
This does not scale and has been proven over and over again. Print never made much money and they had both subscriptions and advertising together to help.
That's pretty much bullocks... Print media costs massively more for publishing costs vs. online publishing. Same wrt television, both of which handle upwards of millions of customers every day just fine.
I'm not sure what point you've made.

Print media does have more costs, but its advertising also costs more. Print also has subscriptions/purchases to actually get the material in addition to the advertising it contained so they have 2 revenue streams, allowing the advertising to be simpler and not carry the entire cost. And it still wasn't a great business model with many outlets just breaking even.

Regardless, there were never "millions" of advertisers or publishers like we have in digital which is why selling direct does not work, unless you only want the top 1% of sites to make any money...

NYT within the past year is claiming 2 million subscribers... now some of those may be online only... but there's every bit of potential they could setup a first party ad system.
They do, it's called native advertising. It still doesn't scale for the 100s of millions of pageviews they get.

And that's the biggest and best publisher there is, which goes back to what I last said: even if the high-end pubs do it, mid-tier and long-tail pubs definitely couldn't handle 1st party ad serving and thus would not be able to make any money.

actually the problem is sharing content that is behind a paywall. nothing against putting that behind a paywal and i definitly would subscribe wsj if i would like their stories / has interests in the content they are posting. (however I don't like such stuff) I've read some articles of wsj that where accessible through google (pretty controversial). also mostly I watch hn and click on a link to read the comments first and then I know that I mostly could ignore certain links.
The problem is the business model.

All of these behaviors 1) are hostile to high-quality accessble content and 2) are fundamentally driven by advertising (or its failure).

Information is a public good. This doesn't mean it costs nothing to produce (the contrary), but high-quality public information has a high positive externality to the common wealth (that is, the commonwealth), and is essential in a democratic society.

Some see the solution as micropayments. I disagree strongly, see Nick Szabo, Clay Shirky, and Hal Varian as to why not. Superbundling might work, but the ultimate superbundle is a highly aggregated, means-based fee. Either a direct tax (income or wealth-indexed) or bundled into broadband services. The former strikes me as more equitable, the latter as more viable.

The costs are low. A $100/year average fee per $30k in income, charged only among OECD nations, would cover all online advertising revenues. A $500/30k rate would cover the entire advertising industry. ($30k is chosen as roughly median OECD per capita income.)

(Note: corrected from earlier figures which applied to worldwide population: $20/person and $100/person would cover everything, if paid by everyone, though I see an argument for progressivity applying the tax to developed nations.)

The distortions, perverse incentives, inefficiencies, privacy invasions, security risks, and more, of the present model simply aren't worth it.

HN's policy is that paywall-based stories are acceptable. I disagree strongly with this view. The WSJ is trying to play both sides of the subscriber-only / viral content game. And are providing increasingly slanted, unreliable, Murdoch-tainted content to boot. I view it as barely above spam.

There are other sources, they should be allowed and encouraged. WSJ should be blacklisted.

And yes, I absolutely would like to see an alternative publication financing model pursued. I'm not holding my breath. But supporters of the model I've proposed include Richard M. Stallman of FSF and Phil Hunt of Pirate Party, UK, among others.

Previously, on global advertising spend, publishing models, and per-capita costs: https://plus.google.com/104092656004159577193/post /adKy4gM317c

Several posts on content syndication, and inherent problems of information goods and markets: https://www.reddit.com/r/dredmorbius/search?q=content+syndic...

Hal Varian (UC Berkeley economist, Google's chief economist) on information goods and markets: http://people.ischool.berkeley.edu/~hal/Papers/japan/index.h...

Since the government doesn't currently pay for Media, how should journalists currently get paid?
I've addressed that in a few articles, posts, and comments, though nothing particularly comprehensive.

One option would be similar to how performance rights organisations (https://en.wikipedia.org/wiki/Performance_rights_organisatio...) operate in music and broadcast based on mechanical copyrights.

Another option would be to identify who the most valuable authors, reporters, composers, performers are, and provide a stable annualised income based on this and related elements.

In the UK, broadcast taxes are paid to the BBC, which funds creatives.

The details of how specifically individuals are paid isn't so important as the fact that they are, in a manner sufficient to provide the information and media necessary for a vibrant culture and society, as well as polity.

If you look at Adam Smith's discussion of compensation for labour, he comes up with a fivefold set of factors, on top of the basic fundamental imperative that wages be sufficient to live on and raise a family.

Somewhere in there lurks at least one successful formula, I'd suspect.

(I'll see if I can't find some relevant links.)

I've updated parent post with links. The G+ and Reddit set contain suggestions on actual mechanism, though I'm open on that.
and this is why I like state-run media sources.

99% of the time I'm not looking for things the ministry of foreign policy doesn't want me to know

and it is favorable toward high quality accessible content. everyone concerned about competition and the competition gave you shit that makes state run news a breath of fresh air

> 99% of the time I'm not looking for things the ministry of foreign policy doesn't want me to know

The other 1% is the news that actually matters.

No, lots of things matter.

That 1% is important, and we absolutely need to make sure we have alternatives for that, but the point is that we shouldn't throw out that 99% just because it's not 100%.

The current system is vastly more invasive of reader privacy than the alternative suggested.

The publishing site, numerous third-party metrics providers, advertising providers, ISPs, and any MITM listener has access to the full metadata, at a minimum, and potentially full data.

You'd need to disintermediate via Tor (blocked or disabled by a great many sites -- Cloudflare and Google are particularly culpable here), or other means, to avoid this.

I've made a habit of using Tor extensively for a number of reasons, tab management among them (opening trivial articles in a Tor browser and having them wash out helps, very slightly, in managing my tab explosion in Firefox/Android).

When there is a missing airplane, and I just want to know relevant information? I'll get the relevant information in a straight forward, to the point fashion.

Local news? Straight to the point, no sensation, build up. Just what happened, what it means.

Military skirmish between two nations that the state has no stake in? Straight to the point.

You are concerned about the human rights abuses and political dissidents. If you need that to rationalize why you prefer to put up with privately owned news sources that can't monetize worth shit, then keep it up. They aren't exactly effective at turning people against about the establishment.

Combining the system with a distributed syndication system would disintermediate publishers from readers while preserving, if desired, the ability to compile statistics on actual readership / viewership.

Privacy and metrics.

If you think about it, this is what broadcast media already accomplish nicely.

> But supporters of the model I've proposed include Richard M. Stallman of FSF and Phil Hunt of Pirate Party, UK, among others.

Those sound like great sources on how to monetize your IP effectively.

Who controls this? The Government? All media is now state sponsored is a questionable model.

>micropayments. I disagree strongly, see Nick Szabo, Clay Shirky, and Hal Varian as to why not.

So my initial reaction to their arguments was that none of their complaints matter as long as it's cheap enough. The fact that my water is metered doesn't make me anxious about usage, because it's too cheap to care.

But of course, that's because water is fungible. I have no fear that the water company is going to start sending me expensive mineral water while I'm in the shower to run up my bill.

But the exact same forces that lead sites to ridiculous extremes of obnoxious advertising, are likely to push up their micro-costs high enough that people can't ignore them, and most likely stop visiting.

The only way I can see it ever working is if there were some way to regulate the costs to avoid this. But I'm not sure there's any way to do that for something so non-fungible.

History says it matters.

Make the individual item payments small enough, and quite literally, it isn't worth my time to have to decide on the value. You see, transaction costs matter, and the actual price paid itself is a very small component. https://en.m.wikipedia.org/wiki/Transaction_costs

Again: the amount of money necessary to eliminate advertising completely is all but inconsequential. The frictions, costs, inefficiencies, and negative externalities of fractional-mill per-page costs would be stupidly inefficient.

E.g., I can read ~50-100 web pages/day. That's, roughly, 36,000 pages, tops, per year, or $0.002 per page. Should I really have to decide one hundred times a day if a page was worth reading?

And to what end?

If I'm broke enough that that's a concern, then there's a tremendous social inequity in denying me the information.

If I'm rich enough that it's not a concern, just bill me the 0.33% out of my income tax that it would cost to provide this. I don't care, literally, it isn't worth my time, and I'm subsidising access for the poor in my own country and around the world.

My whole point is that markets and information are exceptionally poorly suited. I've considered your arguments and rejected them.

Water, incidentally, is another case where the infrastructure, that is, fixed costs hugely outweight the marginal costs of delivery in most cases.

>Should I really have to decide one hundred times a day if a page was worth reading?

No, and the argument I was suggesting is that if the costs were low enough, no one would really need to make those decisions, any more than they need to decide whether running a glass of water is worth the cost.

>If I'm rich enough that it's not a concern, just bill me the 0.33% out of my income tax that it would cost to provide this.

Why? If a micropayment system worked well enough that no one cared about the prices, what reason would there be to involve the government?

(Not that I'm averse to such a scheme, I just think you're making a strange leap of logic here.)

> I've considered your arguments and rejected them.

It feels like you just read my first sentence, and reacted to that, despite the fact the rest of my comment argues against that position, and ultimately agrees with you (although from a slightly different direction).

Then you're creating a per-page accounting system where you still don't need one, with all the tracking and privacy invasion that entails.

Measure share of readership, and you're largely set.

You're re-iterating (best I can tell) earlier arguments, made and dismissed. HN's "only show the immediate post being replied to", particularly on mobile, makes referencing your GP comments all but impossible, but again: I and others far more intelligent than me have considered and rejected the micropayment arguments and pointed out further inequities they ignore.

Well HN isn't just one person so yeah you're going to get different responses.
Yes, you're right.

And if a community of educated, well-informed, and frequently like-minded technology professionals can't come to a semblance of consensus about this matter, don't be surprised if the day comes when publishers -- with an audience far more diverse in opinion than HN -- just throw up their hands and do all three: ads, paywalls, and click bait.

I don't mean to point fingers, as much point out an unfortunate reality.

I work for a publisher. I appreciate quality content. This is a rehashed press release, I shouldn't be expected to pay for it.
Facts are cheap/easy but its still content and you should be expected to pay for it (either via ads or subscriptions). Nothing is free.
In the case of a press release, it's paid for by the company releasing it to the press, so end recipients shouldn't also have to pay as if they were getting an originally produced story.
It's a problem with the model. It's far too much for subscribing to a single site; especially when it's linked to from social media (HN) where it's required to have access to participate in community discussion.

HN will pay for content, but they won't pay under these terms.

I've submitted that as an alternate.

Vote that up and flag this if that's the way you feel: https://news.ycombinator.com/item?id=11592904

(I'm well aware that HN doesn't condone flagging paywalled content. I disagree strongly with that stance, particularly as the paywall repeatedly becomes the primary topic of discussion.)

I'm also more than happy for you to be credited with the submission, though it didn't happen to be submitted already at the time I posted it (I strongly suspected it would have been).

> if that's the way you feel

Everyone who feels good about paywall articles, raise your hand.

crickets

maybe HN could auto-correct...
(comment deleted)
so i guess my super super slow free-shipping orders (non-prime) from amzn as a result of their curbing capex?
You know what this means, right?

Jeff Bezos will increase investment into AWS and Blue Origin, while not investing anything into improving engineer work-life balance.

Seriously, at least cover MSP and transit! Even partially covering a transit pass would be nice!

Sincerely, an Amazon Vancouver SDE I

> Seriously, at least cover MSP

Seriously? That's only what, 60 a month to cover it? Do you guys at least get vision and dental?

Here in Seattle they give you benefits (I only used dental, it was fine for normal treatment) and an orca card (transit pass).

Transit pass is partially covered as of last year in Vancouver. Unless something's changed, you might want to double check on the benefits for Vancouver. If I recall correctly you get approximately 80 dollars give or take.
I don't think the work-life balance issues at Amazon revolve around whether they pay for your transit or negligible medical fee
Woah you're in Vancouver? If you don't mind me asking, what does the pay package look like?
The most interesting tidbit for me is that Amazon would have reported losses without AWS.

The AWS business generated $2.6B in revenues in the fourth quarter, which is only 9% of Amazon's total revenues of $29.1B; but operating income for the AWS business was $607M, or 118% of company-wide net income of $513M.[1]

Without AWS, Amazon would have reported losses!

[1] http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9N...

Which means that they're using profits from an unrelated market to undercut and push out / destroy any competition in their storefront market. Not a good situation for future competition.
(comment deleted)
How to you reach that conclusion from those numbers?
Because that’s literally what they say?

Amazon’s physical stores and Amazon’s online retail business are not profitable. They are selling below price. This is per definition predatory pricing.

They are only able to offset this because AWS is making more profit.

This is evident if you just look at the numbers.

Perhaps it's correct but it's not evident from the numbers.

The numbers can be acceptably explained by expansion, which Amazon does all the time. For example Amazon is just investing in a new retail business in India.

Investments don't actually diminish profits. If Amazon buys a house for 10mill$, their profit doesn't change by a cent, because while they may have less cash, they also own a 10mill$ building.

The value of investments is spread out over whatever is reasonable for the category (i. e.: Building: 10 years or so, car: 3 or 4 years, javascript framework: 8 hours).

True for the building itself, but is buying office space and warehouses a large part of setting up a new business? I'd have thought that market surveys, planning and negotiations, hiring the initial staff etc would dominate.
It depends – you can generally expense anything you buy over it's useful lifetime, so if you aquire a patent for your new project, you'll spread that investment over the next few years. If you hire an architect, it's part of the building's cost.

The only exemption I've ever encountered is intellectual property that you create with employees – there'd be just too much uncertainty over it's value so you suffer those costs immediately (which can be a good thing, it lowers your taxes etc).

But yes, if you're investing a lot there will be some negative effect from the beginning because everything starts losing value immediately and revenue only starts later.

Note that all this is completely removed from cash flow. You'll have to pay for all the stuff you buy – usually pretty quickly. There have been companies with excellent balance sheets which still ran out of money because all the value was in illiquid investments. That's why companies often prefer to lease cars or machinery: it gets the different cycles in sync.

(all based on the laws of my small European country. Not endorsed by GAAP. Lack of acronyms caused by inexplicable loss of the relevant vocabulary once I got away from all the CXXs)

Well said. Though here in the US a significant (sometimes) portion of employee costs lies in stock based comp. And many times, co's present non-GAAP numbers that strip out this part from their income (which inflates their income, and sometimes that can be the difference between making a profit or a loss). The stock based comp gets reflected in increased share count over time, so it will show up in the stock price, just not in the presented non-GAAP income.
Yeah, but paying money to new stuff which don't yet produce profit would make profits less, wouldn't it?
Investments do usually diminish profits in the year in which they're incurred. If Amazon buys a $1000 laptop for one of their engineers, in the middle of the year, and they depreciate it over 2 years, then $250 will be charged to the depreciation account in the current year, reducing profits by that amount.
That's not the definition of predatory pricing. Predatory pricing is setting prices low in the short term, in order to drive out competitors, so as to increase prices through market power. It has nothing to do with the level of overall profits. Profitable businesses can engage in predatory pricing, and unprofitable businesses can charge high and uncompetitive prices.

Your other point about AWS profits allowing losses in the retail business is incorrect. Profit is an accounting measure. A business with positive (even large) profits can run out of cash. A business with negative profits (losses) can operate for a long time, e.g. if it has raised cash by issuing equity or bonds, or if it is generating cash by selling off capital assets.

So does Shopify or other e-shop services run on AWS? If so, doesn't that essentially mean Amazon's competitors are paying Amazon to undercut them?