Why are we allowed to modify or delete old data?

This makes Beeminder to lose all its credibility when I show other people my Beeminder since I can just backdate and input whenever I want to make the Beeminder look good.

1 Like

AFAIK backdated data won’t reverse a derailment that’s been processed, so you can’t hide going of the road using that method.

Beeminder is ultimately an honor code system - it doesn’t have much in the way of mechanisms to prevent cheating, other than autodata integrations. If you can give a more specific description of the sort of data you’re tracking and what your goals for accountability are, folks here may be able to help you find the best way to achieve those goals.


We also have the weasel-proofing checkbox, though at present that largely affects how we interact with you post-derailment.

There was discussion elsewhere about whether manual data entry, data editing, etc, should be restricted on weasel-proofed goals.

But as has been said, there is no way that we can anticipate and prevent all the ways in which weasels are going to weasel. At best we can put in some guardrails to help our nobler selves keep on improving and achieving.


My question is simply what I stated in the OP. Why are we allowed to modify or delete old data in Beeminder? Why am I able to modify/delete data I entered one month or a year ago?

If one is able to modify or delete old data, then showing my Beeminder data to other people doesn’t have much credibility when old data is non-modifiable or deletable. That was my concern. NOT because I want to cheat the system, I am confused to why you would think that. If I want to cheat the system, I wouldn’t even be asking this question, since being able to modify old data enable me to cheat. The very reason why I even ask why data are modifiable is that I am asking why are we able to cheat.


Making old data entry non deteleable and modifiable would help to prevent cheating…

1 Like

What’s the use case here? I’m trying to gauge the value/priority of implementing this as against other improvements that we could make. (As always, the more people that ‘like’ your post, the more development priority it should get.)

Of all the people I’ve shown my graphs and data to, not one of them has ever voiced a concern that I might have manufactured the trail of evidence. I guess maybe that credibility comes from me, rather than from the system.

I agree that there may not be any need to be able to edit old datapoints. (Counter-examples welcome!)

For my own goals, edits tend to happen only for the very recent past. Most often, it’s to fix something that I’ve mis-entered using the form or the app, or which one of my scripts has entered for today because I forgot to enter it yesterday.


In nearly five years of using Beeminder, I’ve pretty much only used the ability to edit old data for the purposes of correcting data entry mistakes. Generally I notice those mistakes within 24 hours, but in special cases (for example, I’m on vacation with only my phone for internet) I might go a week or two before I notice and correct something.

I’ve never had reason to edit data older than that, and I’m hard pressed to imagine what the motivation would be for anyone to do so. There are many other options for cheating which are more straightforward and take less effort. Which means I’m having trouble seeing why it would be worth implementing a mechanism to prevent users from editing old datapoints. The current solution (every data point is editable) seems to me like the least-effort approach to allowing the user to correct mistakes. A simpler solution would be to make all datapoints uneditable, but that would mean a lot more support would be needed to correct all the “oops, I put the wrong date/value/comment” errors.

So essentially, the direct answer to your question is “Because doing otherwise would require more work, and there hasn’t been a compelling reason to do that work.”


Yeah, exactly, you should be able to edit for maybe 36 hours? After that, the data becomes permanent. This would give your Beeminder much more credibility.

1 Like

I was going to put Beeminder link next to my linkedin, but after discovering that all old data are editable. No thanks. I will not be promoting Beeminder to my friends and colleagues. It is embarrassing.

The ability to modify old data points is like giving past financial reports where the past revenue can be changed at will. Zero credibility.

And if something has little credibility, there is no point making it public. At this point, Beeminder has little public display value for goal tracking. The only redeeming point is that you can not back date and add data, but one can still gauge up (or down) the total count by change past data.

You give a certain cut off date, like 3 days, then the data becomes permanent. Beeminder support should have no business in changing people data anyway.

1 Like

I think I can imagine the sort of use case you might be encountering: counting time/output/etc of something related to professional development. Maybe time using a software or something similar? Even if this isn’t exactly your situation, that’s an interesting point.

If I were a recruiter looking for that sort of information from a person’s LinkedIn, I suspect I’d want something I could independently verify (a certification, a link to a github account, a portfolio of work completed, etc). Beeminder doesn’t particularly support independent verification, beyond an autodata graph tied to another independently verifiable service. To use your example of financial reports, the reports themselves are only compilations of data tied to an outside reality: actual transactions, which each could be independently verified. A financial report is only credible if you have reason to trust the person preparing the report. Even if their software prevented them from editing past values, they could still have faked values in the first place.

As always, the more information you can provide about the exact details of what you’re attempting to do, the more likely it is that folks here can brainstorm a solution and/or the feature can be implemented in a way that serves the largest number of users.

EDIT: I just reread your post and saw that you said “next to” your LinkedIn, not on your LinkedIn. Who is your intended audience, and what information are you intending to convey? What sort of credibility are you intending to build? There’s probably a way for Beeminder to help, and users here can help you brainstorm like we did for this person tracking workouts.

1 Like

To give a counter-example to the idea of cutting off edits after a day or two, I just noticed a duplicate data point on one of my graphs that I entered 8 days ago and needed to delete.

(Consider this another vote in favor of graphs that can be toggled among several user-selected display options, one of which is “Past X days.” I would have noticed the duplicate data point much more easily if I could look at a “past two weeks” view while also saving the custom graph settings I currently have.)


I don’t think Beeminder really positions itself as something to provide public credibility. It’s a personal tool: you can add fake data every day, or edit all your past datapoints, or add a bunch of fake backdated datapoints, and craft an elaborate web of lies! You can probably even weasel out of all your derailment payments! But all you’ve accomplished is wasting your own time, lying to yourself instead of actually working toward the goals that you could have been Beeminding to make progress on instead.

Given that I often need to edit or backdate data to make my graphs accurate, including data that’s days or even weeks old, it would really break Beeminder for me if we were locked out of editing datapoints, without making the graphs any more “credible” to outsiders because it still wouldn’t actually be Beeminder’s job to prevent you from entering a bunch of fake data in the first place.


I know this is an old thread but I still find the title relevant despite the content diverting to the credibility of beeminder… so I find the ability to modify/delete data within 24 hours or 36 hours (or some amount but not forever) a good idea but I think being able to delete old data would need to be longer especially in the case of “do more” goals that are auto sourced. As an example I had to weasel proof a goal for doing so many miles on my bike but I would modify a datapoint multiple times to give myself a false buffer but now there are times where beeminder or zapier wigged out and screwed up some data points and now I have to contact support to remove the data points.

1 Like

But again, the core scenario seems to come down to whether you personally need the additional protection of weasel-proofing, or not. Contacting support seems like a good amount of friction to include in such cases, especially if the ‘wigging out’ is on our side of the fence…

1 Like

I can understand being able to edit data points for a manual goal (at least for a certain period of time), but I don’t understand why you can edit (at any time) data points for auto-integration goals like Duolingo. To me, the benefit of an auto-integration goal is that everything is done automatically and you can’t lie. Having the ability to edit the data point negates the whole point of it (well, aside from saving time on data entry).

I had some success with Beeminder in the past, but then I eventually fell behind on several goals and started lying so I wouldn’t derail on everything. I was hoping to create some auto-integration goals (or whatever you call it) to take lying out of the picture.

1 Like

What oulfis said. Even with integrations you frequently have to write and debug code to make the data meaningful.

You may want to look into Boss As A Service - they require corroboration for your data.


A good subsection of the autodata goals do in practice prevent editing the data: they overwrite the previous data, so even when you do try to weasel out and edit in an extra 4000 steps from Fitbit, for example, Fitbit will just overwrite it and you’ll derail anyway.

(And if it would help to be embarrassed about being caught, well: support is all handled by humans and this kind of behaviour is easy to spot. We notice this stuff! You can edit datapoints, but you can’t stop Beeminder autofetching the lower amount you’re actually at. We can also see when people write things like “fake” or “didn’t really do it” in the comments; if you’re not weaselproofed, we typically won’t take note of it, but we see it and as all humans do, we have our private opinions about it.)

If you’re having a big problem with lying, though, that’s the entire point of the weaselproofing feature. You probably won’t like us very much when we refuse to cancel a charge due to inadequate proof the one time you aren’t lying, but it is a really important step if you can’t trust yourself to be honest, and need Beeminder to enforce it. I’ve always found it worth the extra hassle.

Honestly, though, half of what keeps me honest has always been keeping in mind that there are humans behind Beeminder. If everyone lied, they wouldn’t get paid for the work they do! :scream: (Though I says it as shouldn’t, now I’m one of the folks who needs to get paid.)


That is a good point, which I actually just realized a few hours ago. One of my goals synced just after midnight (when I was asleep), so if I had lied the day before I would have derailed anyway.
So that pretty much resolves my issue.

About the rest of your post, I don’t like lying, but unfortunately I find it pretty easy to lie about my data points (at least if I feel depressed and overwhelmed at the time), because it just feels like lying to a computer. I doubt I would lie to the Beeminder support team after I derail because I’m talking to an actual person, but I could just prevent myself derailing in the first place by entering fake data. So I don’t think weaselproofing would help in that case.

But like I said, my issue is resolved now, so it’s not a problem.


My OCD nature makes that impossible for me. Having inaccurate records bothers me much more than lying does. It doesn’t drive you crazy if your records aren’t accurate?


I think I got two things going here that prevent me entering bogus data.

  1. I am very aware of the long and painful journey of failed attempts to overcome my akrasia. When tempted to enter wrong data to not derail so far I have always felt like Beeminder is actually my last resort. If I cut this branch while standing on it I don’t know what else there is that can fix my willpower. And I am SO sick of akrasia. And ultimately I know I betray myself. If I’m not honest to myself, whom can I then be honest to?
  2. Maybe a bit far fetched but it plays into this trust theme: I have never worked at a place where employees had to clock in.
    I have only ever been exposed to what we call “trust based working hours” where the boss trusts the working hours written down by her employees.
    In fact I have never used a clocking in machine in my life. I don’t even remember ever consciously having seen one.
    Maybe that’s just a “Europe” thing. But here this is quite common for white collar workers. So after doing this for a few years this buried itself into my mind very very deep. And I feel like this helps with Beeminder maybe.

Anyway I am a bit surprised nobody brought up point 1 before (or I just haven’t been around here long enough).

1 Like

How about adding hashtags to old data to enable some data mining deep in the past?