Hacker News Reader

Comment by jbl0ndie | Hacker News Reader

I used to use GnuCash but I never quite found it gave me benefits enough to justify the enormous amount of time required for the data entry.

If it was possible for an open source application to support Open Banking (it isn't), that might now be different.

https://lists.gnucash.org/pipermail/gnucash-user/2020-March/...

View on HN

In the US, every bank account and credit card I've found will let you download OFX or QFX files of transactions, and use the GnuCash import features (to match up with transactions I've already added, and to add any I've missed).

There's also a way for GnuCash to sync data with some financial institutions automatically, but I prefer to save the data interchange files long-term anyway.

Sounds like this UK Open Banking isn't entirely an open standard for open interoperation, but maybe a little closer to a cabal?

> UK Open Banking isn't entirely an open standard for open interoperation, but maybe a little closer to a cabal?

That's correct, "open" is a sham, there's nothing open about it. You need to either have regulatory approval or piggyback off a gatekeeper's (TrueLayer, Plaid, etc) regulatory license to even get access, and when you do, you realize how terrible the data quality is and that it's near-impossible to build anything serious on top of it.

The co-founder of a business I consulted for previously recently wrote a nice summary of the issues you'd get if you wanted to embark on that path: https://www.adamjgo.com/p/the-open-banking-non-standards

Unfortunately USAA has disabled OFX and QFX. I hope this doesn't become a trend. I can't believe getting programmatic access to finances can be so difficult in this day and age.

Oh that might be what affected me in Canada.

I was using Mint connected to my RBC account and loved it, but it stopped working forever about 2 years ago.

Now I just do it all on spreadsheets which isn’t as bad as I thought, but Mint was still nice and actually pinged me when I overspent in a category. Now I have to remember (taking away cycles) that I can only spend $X, Y times a month for lunch, beers etc.

You can’t talk to their ofx server or can’t export qfx and ofx files ?

The latter would be terrible. The former seems to be more and more common.

Has anyone asked them to put it back? It might be easier for them to revert sooner rather than later.

Yes people have asked. One they removed access they rolled out a Quicken partnership, so I assume this is some decision made by management as part of a business deal. But I'm only speculating based on the timing.

I bet its a new plan for revenue extraction. quietly remove the feature, then upsell a "premium package" that includes exporting your own data in that format.

Same experience here and with all due respect to their work. I could understand accountants using it but for the general public it forces you to use accounting concept that unnecesary (could be done internally) for managing your personal finances.

In another context this also happens with CRMs where marketing and sales people end up using spreadsheets and from time to time uploading information to the CRM with some push.

> for the general public it forces you to use accounting concept that unnecessary

Strong disagree on the unnecessary part, as far as I know there is one correct way to count money and a billion wrong ways.

Fortunately the correct way, double entry accounting, is actually fairly easy. People were already using this method 400 years ago. It also scales down very well to the level of an individual and scales up to the level of a megacorp.

It's a skill that takes 30 minutes to acquire and pays dividends your entire life.

It took me several years to learn, and I still encounter transactions from time to time that take me one hour to come up with the correct splits.

Not everyone has the same level of financial education, nor the same intuitive mental model around what's credit and what's debit, and what goes where.

This is not my experience, I know accounting but transferring that knowledge (and interest) to people that just want to simplify the registration of a basic movements in a seamless way makes them to look at other tools. It is the UX at the end beyond if you are right on the concept.

ActualBudget (https://actualbudget.org/) supports open banking via GoCardless (https://gocardless.com/). GoCardless (previously nordigen) deal with meeting the Open Banking standards and connecitng to all the banks and provide you with a fairly straightforward API for it.

It looks like automatic importing of transactions isn't supported yet[0]: "There are a couple ways to get transactions into the system. For now, automatic transaction downloading is not available but it will be available very soon."

What I really want is some system that lets me just click "Import", and it goes out to my various CC/bank sites and pulls in the data automatically. Or, imports on a schedule 2x/day or something. The thing I want to avoid is having to log in/download/import these files manually from like 7 different sites. I want that because I handle this stuff on a random basis, when I have a few free minutes, and the time it takes to get my records up to date the manual way is about the same amount of time I have overall to do it. And I find reconciling everything on a fixed schedule, say at the end of the month, is much harder than doing it on the fly, because a lot of times I will not remember what a charge was even for if I wait 30 days to process it.

[0] https://actualbudget.org/docs/transactions/importing

I think the docs need updating but you can auto sync with banks https://actualbudget.org/docs/advanced/bank-sync

Why do you think open source banking is impossible?

Open Banking and open-source client-side software is a poor fit (btw, the "open" is a sham, there's nothing open about it).

Open Banking is implemented as a set of REST APIs using oAuth offered either by the banks directly (if you have the necessary regulatory approval) or by a gatekeeper such as TrueLayer/Plaid/etc (and I believe you can piggyback off their license, so no regulatory approval necessary).

Consuming oAuth APIs relies on guarding a client ID & the associated secret which would be impossible in an open-source app (or any client-side app for that matter, since extracting those is merely a reverse-engineering problem), therefore out-of-the-box Open Banking support is impossible.

However, nothing prevents them from developing the integration and let the end-user provide their own client ID/secret from their own accounts, and I believe most of these Open Banking providers have a free dev plan that can nevertheless connect to real accounts and would be enough for this.

In Germany, we have HBCI/FinTS, it's basically a standardised protocol to do "home-banking", and it works really well. I use an open source Java app, Hibiscus [0], and it connects to all my accounts and downloads the transactions. I can also make transfers in the software and they're pushed to the respective banks.

It's almost what I imagine client-server Open Banking could have been.

There's also the EBICS protocol, but that is more for B2B, server-server communication I believe.

[0] https://github.com/willuhn/hibiscus/tree/master

> It's almost what I imagine client-server Open Banking could have been.

That implementation leaves no room for gatekeepers or other busybodies so there's no way it would be in this day and age.

In the EU and UK, banks are required to impose onerous security and authentication requirements on third party software before granting account access. The rules are designed to protect consumers from insecure or malicious software but unfortunately a side effect is that open source projects are quite unlikely to be able to comply. The most we can hope for is usually that a bank lets you manually download statements as OFX. HSBC does that for example but some other banks/ credit card providers only give you QIF or CSV.

One commenter mentioned aqbanking, which is open-source and can navigate various kinds of 2FA shenanigans via the FinTS protocol. For example, it communicates to my card reader via USB, and relays the resulting TAN back to the bank server.

That thread seems to be attacking the wrong question. The controls are on people, not the software. Individuals should be getting access, and gnucash should just be implementing the protocols for access. Gnucash shouldn't have any secrets in it in the first place.

Not that that would change anything, I doubt banks would give individuals access either.

Reply on news.ycombinator.com