Hacker News Reader

Comment by zzleeper | Hacker News Reader

Devils advocate: Suppose your stats are biased by a % of bad listings (fake, bad, etc.) that violate some rule. Would you want to include them in your numbers?

If having the correct statistics (i.e. excluding these bad listings) was important for regulatory or marketing purposes, then fixing this before the report is part of the fiduciary duty of the firm with its shareholders, so no surprise here.

View on HN

Aside from what other posters have said (a better way to do this would be to note in the report the exclusion of certain data, and why it was excluded), the way they purged the listings is particularly damning:

"No similar event took place in other cities in North America or elsewhere."

If this were them legitimately cleaning up their listings, and not misleading regulators, you'd imagine they'd have taken similar action across their entire database. The fact that they only "corrected" their NYC listings before releasing their NYC report is rather telling, assuming that the data presented here is correct.

Generally speaking, altering your statistics to report to regulators will open you up to a lot of potential legal trouble as the regulators might well decide to take the view that you're deliberately giving them misleading information.

If you need to massage the date to represent what you believe to be a realistic view of your data then you still need to start off with your 'bad' data and explain why you don't believe it's representative and present your arguments for altering your data along with your 'clean' dataset. That way you can't be accused of misleading regulators and you'll have best opportunity to present your arguments for why your initial dataset is 'wrong'.

Your fiduciary duty to shareholders almost certainly includes not making yourself liable to having legal action taken against you by the government.

If regulators are investigating your business, they are already 'taking legal action' against you.

Now regulators can and in many cases often choose to believe that companies act in good faith when they clean up their data profile before handing it to investigators.

Without clear data, regulators can also assume that a company is trying to obscure the truth by throwing everything and the kitchen sink into a data set while internally only using a subset of the data because it knows X part is patently false.

While investigations generally could precede legal action, they don't always lead to it. An individual could be investigated for a serious crime and later be ruled out by said investigators as new evidence comes to light.

That being said, this should lead to legal action by NYC.

Note that this article is a reply to the statements made by Airbnb at [1]. Specifically claiming "95% of our entire home hosts share only one listing" when this is true only because you modified those listing days previously (to remove said multi-home listings) can't be chalked-up to naive data clean-up.

You present no evidence that the purged data wasn't listings that previous made money for Airbnb - there's no evidence it was "bad" in any way except making them look bad and the chart in the article seems to show it was corrected only so long as they needed to look good.

[1] https://www.airbnbaction.com/data-on-the-airbnb-community-in...

I tend to disagree with you. This looks shady, unless they declared the fact that they were trimming their listings prior to presenting the data.

The report is supposed to present the business as it is and has been over recent time. It is very difficult to justify making a whole lot of adjustments to ones business and then try say "this is the way our business looks - honest".

My thoughts, too.

Though the question is why haven't they removed these earlier? Were they waiting to do it? Or was this just a result of incompetence followed by a panicked response?

Could be it never mattered before, because no one was looking, and now someone is looking.

> Could be it never mattered before

If people could rent these properties before they were removed, how could it not matter?

I'd argue that's a form of incompetence.

Because they were making good coin from those listings

Except they only did this in New York. Only if they had gone ahead and did this as a global crackdown, that argument might fly.

There are better ways to exclude the bad listings from a report. The generally accepted way is to include them at the beginning and explicitly excluding them for a number of stated reasons.

This does however give an image that Airbnb is not perfect.

The thing is they aren't just "fixing the report." They realized they had to change their business model, and apparently purge a significant number of customers (who they must have known to have been in violation of NYC legal code up to that point).

To avoid, you know, getting caught with their pants down.

Seems reasonable.

Does that mean that cleanup basically never happened beforehand though? Since they got rid of so many in one go over that one month.

I understand for smaller corps, this sort of cleanup doesn't exactly make money. A bit backseat CEO, but for a company where compliance is important for its continued survival, I feel like that should be the full-time jobs of many people at the company to make sure there isn't a bunch of terrible listings.

Reply on news.ycombinator.com