Hacker News Reader

> > It’s natural to assume that service providers like Ancestry will be good custodians of our data,

> No, never. Not Ancestry or anybody else

I respectfully disagree: it is natural for people to assume that companies won't callously discard their customers’ data. If it wasn't natural, we wouldn't need to have so much effort going towards education about the terms of service to watch out for and how to make your own copies. Too many people have a mental model for corporations which grossly exaggerates the level of stability and long-term planning.

I agree. It is "natural" to assume that they will be good custodians but it isn't "correct" to assume that.

Ok, you have a point. Trust is natural (think about the attitude of little children) and education goes like "don't trust strangers". Hoping for the best often is a sub optimal strategy. I was starting from that assumption.

Trust isn't for "little children." Trust is what makes the world go round. I trust that my bank won't close overnight and steal all my money. I trust that the food I buy is safe for human consumption. I trust that random people on the street won't attack me because they feel like it.

"Hoping for the best" (or perhaps "trusting that people won't screw me over") is actually an extremely good strategy -- far better than not trusting anyone with anything. I can hire third parties to cook my food, do my taxes, handle my banking, even babysit my children after a background check (where I have to trust the background checking company) and a brief interview. If I couldn't trust people, I simply wouldn't be able to function in the modern world.

So I have a company, who I have paid to store my data. The company has been good so far, I use their services frequently, and I have every reason to trust them (they've had my credit card info, and my family's for years, haven't screwed me over, haven't lost any other data in all this time...) The company is shutting down, and they explicitly tell me that they will let me download all of my data, including text conversations. I download the data, only to open it weeks later and find that the company has failed to do what they said they would. In this one particular case, the strategy of "trust" has failed. However, in the long run, it's made my life a heck of a lot better.

Reputation.

We pick banks, stores and neighborhoods based on reputation. We trust some of them because they proved to be good and we don't trust others (I bet there are areas of the world and maybe of your city where you won't walk alone in the night.)

We get educated to trust according to reputation.

Personally I don't trust much Internet services because so many of them die young. I always wonder how many years any of the services I have to use for work will stay with me and how I can move my data somewhere else, in the few cases it makes sense (example: you can export Trello's data, but then what?) My customers and I use them because they are so convenient but I'm sure that sooner or later one of them will die in the middle of a project and we will be burnt. I have redmine wikis from 8 years ago. I'll have Trello's boards for the projects I'm working on now in 2022?

But I also don't trust much any "normal" company. I expect that they make their interests first, mine maybe. I buy from them but caveat emptor.

I think you're wrong if you're saying that "it's NOT natural to assume ...". For sure it's natural to assume a company whose sole preserve is digitisation and archive of family history -- which it has been doing for several years for millions of people -- is going to be a "good custodian".

Is that the best way to proceed? Nope, abject paranoia seems like the only way to preserve data -- assuming there is a nuclear war, have you archived your data on a different continent, have you got a plan for retrieval!?!

I like to backup my data to storage that I own too and I try to keep a copy of it on all my devices but while storage is getting cheaper everyday, hardware makers try to limit local storage on mobile devices every way they can to increase cloud business (even remove microSD card support), Google being a good example...

You would think that by now a $650 phone would have a lot more then 32gb of storage.... But even if you have a phone that supports an external microSD card, Google makes it difficult to write to it using a 3rd party Android app (stating security as a reason).

There should be a law that prevent companies from keeping root to themselves on devices that you own.

A service's customer data must persist somewhere. "On the cloud" is a marketing abstraction to fool customers into a false sense of integrity.

Just remember that most cloud services do not do backups, do not have a disaster recovery / BIA plan, do not use ECC memory in network, compute and storage systems. Users must take personal responsibility for ensuring their private data is safe, because wishing is not a winning strategy.

Amazon uses ECC [1] (search ecc)

MS Azure uses ECC [2]

Rackspace uses ECC [3]

While compute servers aren't backed up this is by design (use storage). Storage is indeed backed up and replicated. Also most assuredly there are extensive disaster recovery plans.

The cloud isn't for everyone, but FUD doesn't help.

[1] http://aws.amazon.com/ec2/faqs/

[2] http://azure.microsoft.com/blog/2015/03/16/microsoft-azure-u...

[3] http://www.rackspace.com/cloud/servers/features

While compute servers aren't backed up this is by design (use storage). Storage is indeed backed up and replicated.

And yet, you are referring to pages about redundancy, not backup. Redundancy helps when a disk, server, or perhaps even data center fails. Backups help when your cloud storage provider makes an error and removes your data from redundant storage.

There have been enough incidents to show that cloud computing alone cannot be trusted [1]. Unfortunately, most cloud services are data silos, so you couldn't even backup data if you wanted to.

And then we haven't talked yet about data leaks. Some larger incident than last year's iCloud leak will probably happen sometime. Hopefully then people will realize that client-side encryption is important.

[1] https://news.ycombinator.com/item?id=8440985

For one the iCloud leaks had not one thing to do with cloud computing. Nor did the dropbox one you linked. Those were application level problems, and could well have occurred with on premises based storage (obviously with a public interface for an icloud issue). Let me assure you that bad synch algorithms have caused data loss before the cloud. Applications can fail to backup or record things.

Also I provided only links to ECC proof.

For one the iCloud leaks had not one thing to do with cloud computing. Nor did the dropbox one you linked. Those were application level problems, and could well have occurred with on premises based storage (obviously with a public interface for an icloud issue).

Sure. But the difference is that if you manager your own storage, you know what backups are made, etc. Dropbox claims that backups are made, but these users lost data regardless.

rsync.net uses ECC memory and always has (since 2001).

We do not do backups, however, unless you are paying for geo-redundant storage. If you use single site storage, the idea is that we are the backup.

Since the conversation has drifted this way, let me also point out that we have s3cmd and gsutil in our platform, so if you're paranoid about trusting data to google or amazon (and to some degree, you should be) you can easily import/export to an independent cloud platform (rsync.net).