Comment by nonameiguess | Hacker News Reader

Comment by nonameiguess | Hacker News Reader

Parent Full thread

There is no reason at all for this to be server-hosted. I have of course done this for myself a very long time ago creating an effectively identical feature set using spreadsheets and even put them onto Microsoft's sharing pages (no idea what happened to them, it was such a long time ago).

I'm not saying this isn't a better solution than Excel. You work with what's available at the time. But there is nothing at all about the size of the data or complexity of the calculations that requires it to be centrally hosted. This is financial data and you're not a bank with reporting and privacy requirements being audited by state charter associations and federal law enforcement. No reason for people to trust you with this data. Make this an executable someone can put on their own machine. Don't require a network connection. Don't send telemetry.

Like Frost1x, though, maybe I'm just a minority and you don't need my business. Ultimately, do what the market will tolerate, I guess.

Let me just say that I'm glad people like you and Frost1x are here, fighting the good fight. Ten years ago, I still though the "make an app on the web" thing was a fad that would go away soon, and the pendulum would swing back to native apps. That hasn't happened, and if anything, the practice of just putting it on the web has become the default, which is a little scary.

The web site looks great, and I sincerely appreciate the courage and vulnerability it takes to throw one's project to the "Show HN wolves," but I have to agree that this needs to be a native application (that I can demonstrably run with the network plug disconnected) for me to touch it with real financial information. It also needs to be "better than old-school desktop Quicken" for its use case, in order to un-sticky me from that application and migrate my 20+ years of data from it.

> That hasn't happened, and if anything, the practice of just putting it on the web has become the default, which is a little scary.

The problem is that everyone making something like this wants to have at least seven zeros in their bank account when they cash out after 5-10 years - selling licenses to a desktop app doesn't get you that.

There is absolutely a reason. It's a service and many people (myself included) don't want to download a desktop app to use it.

He's not asking for your SSN or your address. If you want to use a burner email and a VPN or enter fake data to try it out, you can. This hardline approach of "every service should be an open source desktop app that I can run with zero telemetry from my cabin in the woods with no internet" so often found on HN is clearly not the opinion of the majority of internet users.

> He's not asking for your SSN or your address.

They're asking for even worse - all your historical financial data. Why would you want to give some random guy on the internet that?

I'm trying to understand the concern but most of my worry around "historical financial data" would be actual transaction-level information, not rough summary figures to estimate a financial plan.

I just played around with the site myself and I didn't see anywhere that it imports data, plus the tagline on the homepage says "never ask to link your financial accounts"?

Isn't this essentially the same risk profile as like FIREcalc [1], or the NYT Rent or Buy calculator [2], or some portfolio analysis tool [3], are you saying that those shouldn't be online calculators either?

[1] https://firecalc.com

[2] https://www.nytimes.com/interactive/2014/upshot/buy-rent-cal...

[3] https://www.portfoliovisualizer.com

Again, unless you provide some _other_ piece of data (full real name, email, etc) connecting that data back to you, I fail to see the harm. Regardless, no one is forcing you to use the site, or even provide real numbers. If anonymously providing your salary and savings data worries you then don't use it.

How is installing an executable on my machine and then entering my financial info any less sketchy than entering my data into a website? The average user would still have to take it on faith that the data isn't going anywhere, and as an industry we have spent the last 20 years telling people not to download unknown software onto their systems. I think it would be a complete own-goal for the builder of this to switch to that model.

I can install it in a network isolated Qube. Data wouldn't go anywhere.

Sure, many of us can do stuff like that, but the average user can't.

But the default expectations are very different. The default expectation for a web hosted application is that all your private stuff is stored in a database with dubious security, and if your data is stolen as a result the only consequences are a 'whoops, my bad' (at most!) from the company.

It can run client side and store data in secure cookies. Maybe that's what was meant? And then export to a file for backup.

I think many people have gotten out of the habit of installing software on their computers and prefer web-based solutions.

The big problem is: installing a dubious (not saying this is dubious, but lets say anything not from a MNC) program on your computer might lead to a hostile takeover of your computer. But a web-based solution means your data is in somebodies else hand. So one has to pick ones poison...

I’d like to see everything as a web app (a secure sandbox everyone already has installed in their computer), but running with service workers and using indexeddb for persistence. Best of both worlds: you can access everything offline, the data stays local, and the app can’t infect your computer (absent a browser exploit that’d be bad news regardless)

For me I break it down as such: If it's supposed to be "social", I'll use the web based app (ie: twitter, etc). If it is personal (ie: banking), unless it is my actual bank, I want that on my machine.

You can still have it be client side if you're web based, right?

How?

By doing all of the computation of user data clientside, and making all of the persistence in localstorage.

It’s very doable, especially with small data.

Unfortunately, you have to take projects’ words that they do this on good faith as an end user. You’d need to GET the app bundle, and then firewall the tab / disable your network interface to be certain that no data is leaking out. It’s not so easy to control this as an end user these days, and web apps can be quite sneaky.

Maybe we’ll see some innovation in this space from browsers in the future, where an app can identify itself as local only and then be sandboxed by the browser.

> Maybe we’ll see some innovation in this space from browsers in the future, where an app can identify itself as local only and then be sandboxed by the browser.

That would be a really cool 'HTML5' API — let the page do:

  document.addEventListener('DOMContentLoaded', () => {
    // some bootstrap
    document.goOffline();
  });

Which could prompt the user — "this web page wants to disable its own internet access. Do you want to continue? (Refresh the page to reload from server)[OK/Cancel]" — and then it can do longer use the internet & has its own storage which gets wiped before it's allowed back online.

Would allow all the benefits to devs of doing things in the browser (if that's their choice), & end-users could trust that it was truly offline.

Probably this will end up like all the apps which ask for all kind of permissions they really don't need

I Agree very much with the general idea. One addition, though:

> Unfortunately, you have to take projects’ words that they do this on good faith as an end user

The same is true for applications running natively. Of course, those can be firewalled from any outgoing and incoming network traffic more easily. And they don’t feel like and don’t default to using a network connection all the time. But still, they could.

It might just be easier to partner with a respected company or companies that people trust with their money/data already. The biggest one that comes to mind that also seems a bit behind on the visualizations and sleek UI that this has would be Vanguard.

Any thoughts on the comments under the Frost1x post? (e.g. Electron app version, and/or import/export or local storage options for web premium users). Certainly local storage is very doable -- I didn't anticipate how hotly requested that might be. I'll also note that all calculations are actually already client-side.

I like the impression of your tool, but I share Frost1x’s reluctance. However, I was pleasantly surprised to read this up front:

> Free planning and projection tools that will never ask to link your financial accounts

If it were followed up by convincing me that my data is stored in LocalStorage and doesn’t leave my browser, I’d be sold :). I know I’d still have to trust you that you’re actually implementing what you claim, but I think I’d be willing to. After all, the same would be true for an application running natively: those, too, can submit my data “to the internet.”

For me personally, having an Electron app would not make much of a difference. Do I run your code in my Firefox or in a separate process? I’m still running your code on my machine. Maybe browser-based even has some advantage because it is sandboxed filesystem-wise. Not sure whether this holds true for Electron.

Import/export as CSV or whatever format would also be a huge plus for me.

As a last point, I’m not sure I understand the status quo: you are saying all caluclations are done client-side – but the results are stored server-side?

Nice work!

Thanks! A LocalStorage only option seems very doable. With all the interest expressed here, I'll add that to the roadmap. As for the status quo, without Premium no plan data is stored server-side. With Premium, plan data is stored server-side only if the user chooses to enable that feature. Calculations/projections are always made client-side and the results are not stored server-side.

JavaScript (store + compute everything client side)

PWA

I started out using excel and quicken but at some point ended up using beancount, fava and python. There are so many variables like some investments are tax free, others are tax deferred, some are exempt from state but not federal taxes, etc. I didn't dive in too deep to the app, I agree with other comments here that it would be nice if he had some demo accounts with a few different scenarios, like single, family, different ages, incomes and investments to showcase how the modeling works.

Reply on news.ycombinator.com