2009

What washing machine to buy?

We are considering replacing our old washing machine.
Currently we have Bosch front loader that works for us since 2009 (12 years).
Unfortunately, due to wear and tear, this old washer has some internal leaks and smells mold a little bit.

Should we just buy [slightly] newer Bosch front-loader again?

Or is it better to buy Speed Queen?
Speed Queen is very reliable and quick. But because it is a top-loader - it may be not able to wash big blankets. Originally posted at: https://dennisgorelik.dreamwidth.org/211957.html
2009

Tooth fairy

~~~~~~~
https://www.youtube.com/watch?v=prbSQ4rMKhU&t=24s
Now you put that [your tooth] under a pillow, and the tooth fairy will leave you a penny.
...
There is no tooth-fairy, idiot!
Now, clean up that horseshit!
~~~~~~~
From comments:
===
TR0X3N [3 months ago]
His dad waited for hours just to watch his son find horse shit.
Now that’s dedication.
===
Laceykat66 [3 months ago]
Seems like a father who is invested in his son's education.
===


What impact does such "horseshit" lesson have on a child? Originally posted at: https://dennisgorelik.dreamwidth.org/211556.html
2009

The "War on terror" in Afghanistan

https://www.youtube.com/watch?v=ZvZj3u1Eenw&t=2848s
"The marine artillery fire killed three men, an old woman and a small girl.
It wounded five more children.
Even though it had been marines who shelled the village, the villagers loaded some dead and wounded into cars and drove for help to the marines."

Here's one of the wounded girls:
https://youtu.be/ZvZj3u1Eenw

She lost her arm:
https://youtu.be/ZvZj3u1Eenw?t=5718

This is a very sad picture of the "War on Terror" that the US was waging in Afghanistan. Originally posted at: https://dennisgorelik.dreamwidth.org/211362.html
2009

UVXY PUTs

Today is UVXY PUTs trading day again.

VIX was somewhat high last night (up to 23).

In the early session (9:30 ... 11:00 EDT) I bought some UVXY PUTs (and some VIX PUTs).
Then a surprise [at 14:34 EDT] -- VIX jumped to 23.27 and UVXY jumped up to 29.21
So I bought more UVXY PUTs (while they were cheap).

If VIX will go down tomorrow -- these UVXY PUTs will cost significantly more. Originally posted at: https://dennisgorelik.dreamwidth.org/210578.html
2009

Boogieboarding incident

Yesterday night I went boogieboarding.
I rode couple of ~1m waves in shallow waters.
Then I [almost] caught ~1.5m wave. This wave threw me down into ~70cm of water.
After going through these 70cm of water - I landed the sandy ocean bottom (my boogieboard below me).
It re-confirmed that sand is not very soft when you fall on it.

My body still hurts [slightly] in 3 of places [from this yesterday ocean bottom landing].
The hurt inventory:
1) Below left shoulder somewhere around ribcage on the back/left.
2) Around right-bottom ribs.
3) 5 mm length scratch about half-way through my skin close to the base of my right pinky finger.

Nothing serious, but a reminder that waves can be dangerous, especially in a shallow water. Originally posted at: https://dennisgorelik.dreamwidth.org/210316.html
2009

What to tell Aggregator about bots traffic?

In previous posts I explained How bots traffic affects advertising and How to detect bot traffic?

Now we are ready to discuss what to communicate about bots traffic to our advertisers.

Not charging our direct customers (advertisers) for useless bots traffic -- is a reasonable choice if we care about long-term relationships with our customers.
But what to do when things are getting a little bit outside of my direct control?

"Aggregator" case
Let's consider a practical example.
One of our advertisers is a job aggregator.
Let's call that advertiser "Aggregator".
Aggregator gets (aggregates) jobs from other job boards. These job boards are Aggregator's clients that advertise jobs on that Aggregator.
Aggregator advertises these aggregated jobs on multiple publishers' websites.
PostJobFree is one of many publishers that Aggregator uses.
The aggregator may charge advertisers 0.24/click and then pay publishers 0.20/click (20% margin).

So if publisher's traffic is inflated with bot clicks, then in the short-term, Aggregator may financially benefit from that bots traffic.
In the long-term, that useless bots traffic is likely to destroy Aggregator's relationships with advertisers.
So we may expect that reasonable Aggregator would want to promptly identify and discount useless bots traffic.
In real life, business decisions turn out to be a little bit different.

Aggregator's team focuses on making money and tries to eliminate distractions that slow down money making.
So Aggregator's team ignores my subtle hints (I hint that it may be useful to discuss some techniques about how to identify bots traffic).

I think that Aggregator's team may even get a little bit annoyed by my attempt to talk about bots traffic, because the more Aggregator talks about bots traffic - the harder it is to claim that Aggregator knows nothing about that bots traffic.
On another hand, Aggregator may like charging advertisers for all traffic (real users + useless bots), but pay us only for real users traffic (because we do not count bots traffic in our invoices).

I try to not push too hard with my warnings about bots traffic.

What to communicate?
I consider these 3 approaches of communicating bots traffic issue to Aggregator - reasonably practical:
Approach #1: "Keep it quiet"
Silently discount bots traffic and tell nothing about bots traffic to Aggregator [unless Aggregator team asks about bots traffic themselves].
Ignore differences between PostJobFree clicks counts vs [inflated] Aggregator's clicks counts.

Approach #2: "Balanced"
Discount bots traffic and briefly mention to Aggregator that we discounted bots traffic.
Do not try to explain further without explicit interest from Aggregator.
(I currently follow that "balanced" approach).

Approach #3: "Full disclosure"
Be very direct with Aggregator and clearly explain all the dangers of bots traffic and how charging for useless bots traffic may negatively impact long-term relationships between Aggregator and their advertisers.
Explain even if Aggregator does not ask about bots traffic.

What do you recommend to communicate to Aggregator about bots traffic issue? Originally posted at: https://dennisgorelik.dreamwidth.org/210128.html
2009

How bots traffic affects advertising

PostJobFree charges other job boards for job seekers traffic that PostJobFree sends them.
If PostJobFree counts bot traffic as a valid traffic - that may increase advertising revenue quite significantly.
Unfortunately, bot traffic, almost never has any value for the advertisers.
So in 1-3 months most advertises figure out that the traffic is inflated and does not deliver enough value for the cost.

In business it is better to focus on long-term goals and try to keep customers (advertisers) happy.
I want our customers to keep buying job seekers traffic from PostJobFree -- for many years.

That is why we:
1) Put a significant effort for detecting useless bots traffic.
2) Do not charge advertisers for bots clicks.

Still, some decisions about what to do with bots clicks count -- are tricky.
In particular, when I need to decide how much to communicate to the advertiser.
I will write about one such tricky case in the next post. Originally posted at: https://dennisgorelik.dreamwidth.org/209916.html
2009

How to detect bot traffic?

1) In May 2021 PostJobFree started to receive a lot of bot traffic [originated from Microsoft network - probably hosting].
Under "a lot" - I mean about 3x of usual PostJobFree traffic.
These bots come from multiple IP addresses, and that made it hard to identify that bot traffic.

2) PostJobFree has bot detection systems for many years. It successfully detects bots that come from a single IP address.
But if bot spreads the requests to multiple IP addresses, then every individual IP address sends not a very noticeable amount of traffic.
Still, all together, multiple IP addresses send a lot of bot traffic.

3) I found that we may detect these bots by grouping traffic counts into Cnets ("Class C" IPv4 networks).
IPv4 address has 32 bits.
First 24 bits of IPv4 address define Cnet [that this IPv4 address belongs to].

If PostJobFree website counts traffic for individual IP addresses separately - then PostJobFree detector misses many bots.
But when PostJobFree groups traffic per Cnet - then PostJobFree is able to detect [almost] all high-traffic bots (and still is able to correctly count regular users clicks).

4) Bot traffic may seem like a non-event.
Why do I care about that bot traffic?
One of the reasons why I care about bot traffic -- is that bot traffic may dramatically inflate advertising cost.
I will write about the consequences of inflated advertising costs in a separate post.

5) Do you know other ways to detect bot traffic? Originally posted at: https://dennisgorelik.dreamwidth.org/209619.html