FREE! Subscribe to News Fetch, THE daily wine industry briefing - Click Here

USA Wine Ratings

International Bulk Wine and Spirits Symposium

Google’s inability to deal with fake accounts & content makes reviews UNtrustworthy

This is article 1 in a series. Article #2 is:Dissecting a winery’s Google troll attack via multiple anonymous accounts and fake “reviews”

Russian trolls and other fake content creators who have tried to influence U.S. elections and other public policy issues have become hot topics and brought hot-seat grilling to Google, Twitter and Facebook.

And while Google might be working on something aimed at global trolls looking for world domination, it seems to be okay with allowing fake news and reviews aimed at damaging the livelihoods of small businesses … or falsely promoting those who pay.

Indeed, anonymous fake reviewers are a chronic problem that algorithm giant Google seems incapable (or unwilling) of coping with:

Basic algorithm 101

An algorithm is just a series of steps aimed at solving a problem. For example:

  1. If water is dripping from the ceiling.
  2. Then check to see if it is raining.
  3. If it it raining, then call roofer and buy tarp and buckets
  4. If it is not raining, then turn off water at meter and check pipes in ceiling.
  5. If pipe is leaking, call plumber.
  6. Etc.

As an algorithm designer myself, I realize that to create an algorithm, one needs to:

  1. Recognize a task/problem that can be addressed by a step-by-step mathematical process.
  2. Think thr0ugh every small specific of the task and categorize them.
  3. Identify the data needed to feed the algorithm, making sure to include all possible sources.
  4. Create an algorithmic process to address those issues.

There are other steps, but that’s a high-level view.

If Google realizes it has a task/problem with fake reviewers and fake news, there are a number of well-established steps toward creating an algorithm to address the situation.

How Google stacks up (falls flat) on the basics

#1  – In the case of fake local reviews, Google would have to ignore step #1 (problem exists) given the enormous coverage and widespread attention that fake reviews have generated.

#2 – Given its algorithmic prowess in other areas it’s hard to comprehend that Google isn’t really very good at thinking its way through the problem well enough to get started on an algorithm.

But, assuming that is the case, here are a few pointers:

Fake reviews tend to be short in length and lacking details. They tend to focus on extraneous issues and are often personally abusive rather than dealing with details that may assist third-party readers. Fake accounts usually lack profiles and identifying graphics. Those which have posted photos tend to use screen captures of existing web images without knowing that Googles own image search can usually track those down.

The posts by fake reviewers who have multiple anonymous accounts use similar language and sentence structures which can be tracked to the source by forensic linguists using textual analysis.

The textual analysis done by forensic linguists use a process often — note to Google — incorporated into algorithms.

More hints to help Google design an algorithm:

#3 – Google’s got data. Its ability to suck in data is second only  to the NSA. Perhaps better.

#4 – Google’s got algorithm prowess. It’s has proven its algo-talent in myriad ways — from self-driving cars to targeted advertising.

First, it has to matter

Actually, There should have been a “Step zero” — You’ve got to gove a s**t about something before you spend the needed brain cycles on creating an algorithm.

Given the fact that Google has data and the ability to create world-class algorithms, its failure to create a system to detect fake reviews can only result because it simply doesn’t care.

And for an online-based company to require phone calls instead of having an online alert system is further evidence of a  malignant disregard.

On the other hand, Yelp moves on trolls quickly, proactively

Fake reviews still plague the Yelp system. However, there are many links like the three below that demonstrate that Yelp has pro-proactively pursued the problem, acts quickly, and responds to tips about trolls.

Does this mean that Yelp is all warm and fuzzy about businesses that are victimized by trolls?

Probably not.

However,  Yelp realizes that its only real value — and ability to make money — lies in the credibility of its reviews.

Trolls are a threat to credibility.

Therefore, trolls are a threat to the income line in Yelp’s profit-and-loss statement.

The syllogism fails for Google because it has massive and many other ways of making money. Reviews don’t even rise to the financial level of: “would you like fries with that?”

This is article 1 in a series. Article #2 is at: Dissecting a winery’s Google troll attack via multiple anonymous accounts and fake “reviews”