HNHacker News
TopNewBestAskShowJobs

mikpanko

983 karma · joined May 2, 2014

mikpanko(at)gmail.com
submissionscomments
mikpanko··on The future of generative AI is niche, not generalized
Because many people write for virality and clicks and pieces with strong opinions usually have higher reach. “We don’t know” narrative doesn’t sell as well.

There are many people who write deep nuanced articles too, but since they have lower virality, we are usually exposed to the first type of shallower articles and it might seem that there are more of them.

mikpanko··on Substack Notes Launched
In my experience the reality is much more nuanced:

- VCs don’t want founders to own 0% of their company because founders need to be motivated to work hard to make it a success

- % of dilution usually goes down very significantly over funding rounds

- there is significant competition between VCs to fund good startups these days, which can translate to founder leverage

- there are early-stage VCs these days, which don’t pressure founders for quick growth

- founders talk to each other and a large portion of founders are serial entrepreneurs. Reputation among founders matters to VCs

- looking over the longer term of decades, typical funding terms are getting much more founder-friendly

mikpanko··on Substack Notes Launched
“Venture Deals: Be Smarter Than Your Lawyer And VC” is pretty good. Used it when raising a round of funding.

“The Power Law: Venture Capital And The Making Of The New Future” is also good. It tells the story of the evolution of VC over the last 70 years. It is interesting that funding terms seem to be becoming more and more founder-friendly over decades.

mikpanko··on Google CEO on AI, the Future of Search, Efficiency and Battling Microsoft
https://archive.is/d74bi
mikpanko··on Ask HN: How do you test SQL?
Datafold (https://www.datafold.com/) is solving exactly that problem and has integration with dbt.
mikpanko··on Unconventional startup culture by LATOKEN CEO (Russian)
CEO defending his method [1]:

BE PARANOIC - YOUR BUSINESS IS STEALING RIGHT NOW ON ZOOM Or how to survive in crypto and create a scalable product with an effective team of 100-300 people remotely?

Upd. The video can be compared to a torture tape from someone's bedroom. Well, ok, this is "bdsm" with the rules that everyone agreed on and the zoom out button. You need olympic-level bdsm in a startup, that's your only advantage over legacy companies that they can't copy. Did you see the reaction to the video? Can not.

Briefly: New rounds of evolution and products are created by non-standard individuals under the pressure of a threat to survival - open the eyes of the team to it. Build the culture of Olympic athletes with automatic up or out. Describe it and reward with options those who identify free riders who break your rules. Hire and train athletes with case studies, bringing problems and even spanking to the general discussion. Teach them to conflict openly in order to achieve a result. As long as your up or out doesn't work automatically, push as hard as you can to get the result or out so that the free riders don't decompose the athletes. But at the same time, guarantee the execution of contracts, and allow employees to swear at you, but in front of you, and not in the locker room behind the “coach”.

You have to and you can afford it: in web3 channels this is a constant background - in any case, you will be attacked by free riders and thieves, and the public likes it. Think of these attacks as a publicity stunt to magnetize like-minded people who can create the future.

[1] https://www.facebook.com/valentin.preobrazhenskiy/posts/pfbi...

mikpanko··on Nobody optimizes happiness
Is optimizing for happiness a worthy goal? Another approach is optimizing for having meaning in life. As humans we can’t escape some suffering in our lives, but doing it for something we consider meaningful makes the suffering more bearable. Man’s Search for Meaning by Viktor Frankl is a classic book on the topic.
mikpanko··on Shamelessness as a strategy (2019)
Most known examples of “shamelessness” today, as defined in the article, are probably Donald Trump and Elon Musk. The strategy worked incredibly for both.
mikpanko··on Emoji Kitchen
Reminds me of the Little Alchemy game: https://littlealchemy2.com/
mikpanko··on Ask HN: Who is hiring? (April 2022)
Motif Analytics | Founding Designer, Founding Engineer, Founding Data Scientist | San Francisco or REMOTE | Full-time | https://motifanalytics.com

Motif is an analytics product tailored to answer complex business questions through analyzing event sequences. It goes beyond traditional reporting (what is happening to the key business metrics?) and instead focuses on root-causing (why are metrics moving?) and guiding (how can the metrics be moved in the right direction?) [1].

Today, questions about product gaps and opportunities require multi-week deep dives, and often just raise further questions. As a result, many business questions go unanswered and teams rely on hunches to come up with and prioritize features [2].

Motif solves this by providing an interactive data wrangling, visualization and exploration experience to find insights in event sequences - “motifs”. Designers create magnificent art in Photoshop, game developers build powerful games in Unity and engineers architect reliable constructions in AutoCAD. Motif is a kindred power tool for data practitioners.

This approach has been proven at Google and Uber, and is ripe for bringing to the wider data ecosystem.

We are iterating on an MVP with a few early customers, are well-funded, offer competitive compensation and are hiring the founding team.

If you are interested, or you know someone who might be, please reach out to misha@motifanalytics.com .

[1] https://motifanalytics.medium.com/the-missing-tool-for-data-...

[2] https://motifanalytics.medium.com/why-cant-you-pull-data-rea...

mikpanko··on Ukraine Evidence: open data and info verification project about the conflict
Many of us have been watching the horrifying developments in Ukraine over the past month and are wondering if and how we can help. Many are donating money to humanitarian causes but some are also wondering: can we help in a more scalable way through our tech skills?

There is a major problem with getting accurate information about what is happening on the ground in Ukraine. In the world of the internet and smartphones, there is an abundance of video, image and text accounts coming out but they are overwhelming, unstructured and contain a significant number of fakes. This leads to proliferation of misinformation, confusion and the ability of malicious actors to spread their own fake narratives, prolonging the conflict and causing additional damage.

Here is a prominent example (as of March 25). The city of Mariupol is surrounded by Russian forces with hundreds of thousands of civilians trapped inside. The city is heavily bombed and people live in basements for weeks. There are reports of hundreds or even thousands of civilian deaths and of a major humanitarian crisis with the lack of clean water, food and medicine. There is an account of the situation from two journalists, who have recently escaped the city: [1], [2]. They report Russian soldiers being responsible for the situation and provide videos and photos. There are also accounts from pro-government Russian media about Mariupol, for example: [3], [4]. They report Ukrainian soldiers not letting civilians leave the city and providing videos, photos and possibly witness accounts.

Can we apply skills in big data, analytics and machine learning to help?

We believe we can use the power of:

- thousands of people to collect, label and connect pieces of evidence

- dozens of ML groups to run their ML algorithms to establish truthfulness

- data experts to systematize and make evidence widely available and searchable

This can help:

- more people around the world to be better informed instead of only following sources from one information bubble they happen to be in

- governments to better understand casualties and damage

- volunteers on the ground to coordinate their efforts more effectively

- people who lost touch with relatives and friends in the conflict zone to find more about them through image search

- the world to document and preserve information about the big historic event

… and in many more ways, which we can’t think of ourselves right now.

Here is a short link with more details about the project: https://bit.ly/ukraine-evidence.

We are looking for the project leader and 1st ring team members (see the last section of the document for details). Reach out to ukraine.evidence@gmail.com if you are interested and share the document with others who might be.

[1] https://apnews.com/article/russia-ukraine-europe-edf7240a9d9...

[2] https://meduza.global.ssl.fastly.net/feature/2022/03/21/msti...

[3] https://russian.rt.com/russia/article/980673-bezhency-mariup...

[4] https://t.me/margaritasimonyan/10764

mikpanko··on I think US college education is nearer to collapsing than it appears
One interesting idea is replacing traditional student loans with personal funding based on "personal equity", similar to VC funding of startups. Backers would provide funds for individual's education in exchange for a percentage of their lifetime earnings. Obviously, there is a number of questions to think through here, but the incentives of all participants seem to be much better aligned to produce best results for the society. Lambda School / BloomTech and a few other companies are already looking in this direction but haven't broken into mainstream yet.
mikpanko··on The ClickHouse Community
Thank you for the links! Do you happen to know where Facebook’s Scuba, Google’s Procella, Google’s Dremel on Capacitor and Snowflake come on that list of benchmarks?
mikpanko··on The ClickHouse Community
Is it faster for querying than using Presto on top of Hive? The page comparing ClickHouse with other analytics solutions doesn’t list Presto, which is very popular these days.
mikpanko··on The ClickHouse Community
What is the difference between ClickHouse and Apache Parquet or Google Capacitor?
mikpanko··on Show HN: Interactive data visualization of NFL plays (also user sessions, etc.)
Hi all, visualization developer here.

The idea for the visualization comes from my challenges working with user session data in several tech companies (however, it can readily apply to process data in manufacturing and services industries, sequences of plays in sports, etc). Raw events/logs are usually considered hard to work with, so it is common to collapse them into key metrics/dimensions and plot those in dashboards. This works well for basic reporting but not so much when one needs to root cause observed changes/differences in metrics or wants to explore data to find product gaps/opportunities. Here are a couple of very common questions: why have conversions dropped in an updated UI? What do users do when they don’t convert?

Additionally, SQL-based tools, which are commonly used for analytics, are pretty inefficient for querying sequence data. Most table schemas and query languages assume metrics/dimensions data structure, which doesn't readily lend itself to asking questions about the order and relationships between events in the same stream.

With this data visualization, I am exploring two major questions: how to visualize sequence data in an intuitive way? how to intuitively ask questions of sequence data?

To play with the first question, try hovering, clicking, double-clicking, right-clicking, long-clicking and dragging on the Sunburst graph. To explore the second question, read through a few examples of the query language below the graph and then try adding a few queries into the “Rules to filter & modify sequences” widget. This is just a prototype, so can be rough around the edges.

Would love to get your feedback and ideas for improvement! Especially curious about interesting questions, which the visualization can’t help answer yet, and ideas how to add that capability. Would also appreciate ideas for other popular datasets to plot. If you have a business sequence dataset, which you’d love to try with the visualization, please reach out to me over the email in my profile.

Thanks for checking out the data viz and happy to answer any questions!

mikpanko··on Show HN: Interactive data visualization of NFL plays (also user sessions, etc.)
Hi all, visualization developer here.

The idea for the visualization comes from my challenges working with user session data in several tech companies (however, it can readily apply to process data in manufacturing and services industries, sequences of plays in sports, etc). Raw events/logs are usually considered hard to work with, so it is common to collapse them into key metrics/dimensions and plot those in dashboards. This works well for basic reporting but not so much when one needs to root cause observed changes/differences in metrics or wants to explore data to find product gaps/opportunities. Here are a couple of very common questions: why have conversions dropped in an updated UI? What do users do when they don’t convert?

Additionally, SQL-based tools, which are commonly used for analytics, are pretty inefficient for querying sequence data. Most table schemas and query languages assume metrics/dimensions data structure, which doesn't readily lend itself to asking questions about the order and relationships between events in the same stream.

With this data visualization, I am exploring two major questions: how to visualize sequence data in an intuitive way? how to intuitively ask questions of sequence-data?

To play with the first question, try hovering, clicking, double-clicking and right-clicking on the Sunburst graph. To explore the second question, read through a few examples of the query language below the graph and then try adding a few queries into the “Rules to filter & modify sequences” widget. This is just a prototype, so can be rough around the edges.

Would love to get your feedback and ideas for improvement! Especially curious about interesting questions, which the visualization can’t help answer yet, and ideas how to add that capability. Would also appreciate ideas for other popular datasets to plot. If you have a business sequence dataset, which you’d love to try with the visualization, please reach out to me over the email in my profile.

Thanks for checking out the data viz and happy to answer any questions!

mikpanko··on Hemingway Editor
I have built a similar tool to help me edit my PhD thesis a few years back: https://expresso-app.org/. It helped me improve my writing but was also picked up by some English teachers to use as a scalable way for providing initial feedback to students, so they can later focus on deeper feedback in their own grading. So I now keep the tool running for the small community.
mikpanko··on Lessons I learned from the failure of my first startup, Dinnr
Looks like Blue Apron is doing much better with the same idea in the USA - according to CrunchBase they raised $50M in round C several months ago (http://www.crunchbase.com/organization/blue-apron).
mikpanko··on Live map of Russia advancing into Ukraine
I am a Russian coder / data scientist. I am not sure if this particular map is propaganda or not, but I follow Russian, Ukrainian and international news and it seems pretty clear that Russian military is strongly supporting pro-Russian rebels in Ukraine:

- there are numerous stories and videos of Russian "volunteers" with heavy military grade weapons crossing Russian-Ukrainian border throughout the last several months. Some of them are captured by Ukrainian military. Some others are wounded or killed and taken back to Russia for treatment or burial

- pro-Russian rebel leaders often admit to direct Russian military help on video

- just several weeks ago Ukrainian military was advancing fast and getting close to crush the rebels in the last two large cities Lugansk and Donetsk. Suddenly, along with more information about Russian support, the situation reversed and the rebels are on the attack now

- Russia used a similar tactic in Crimea in March: first, suddenly numerous unidentified soldiers appeared on the peninsula and Russia denied they were from Russia. Later, when Crimea was annexed, Russia openly admitted to using "special forces" and rewarded its troops for a successful operation.

Russia may be constantly demonized in the Western media, but there are good reasons for it. Also, just to note, the situation with demonization is much worse inside Russia with extreme lies and propaganda over TV and official mass media.

mikpanko··on Ask HN: Who wants to be hired?
Boston, Relocation, Full Time

Stack: Python, Javascript, MATLAB, SQL, MeteorJS, Git, Balsamiq, Illustrator

Resume: http://bit.ly/mikpanko-resume, https://github.com/mikpanko

Contact: mikpanko@gmail.com

Finishing PhD in computational neuroscience. Strong background in physics, mathematics, machine learning, neuroscience, and programming (but not in large-scale web-programming yet). Led several academic and client-facing software projects, such as a social network for Russian civil activists (http://elections.grakon.org) and a small tool to improve English writing style (http://www.expresso-app.org). Interested in technical product management and also data science and software engineering positions.

← PreviousPage 3 of 3