The comment file got corrupted, so we're trying to figure out how to fix it. In the meantime, the blog is not currently displaying comments. It appears to be storing the comments, however, so I hope we'll get it fixed within a few days.
1 Comment
Leave a comment
Subscribe to Entry
For more info on our research:
Blogroll
Sister Blogs:
Statistics:
- Chance News
- The Endeavour
- Christian Robert
- Revolution Computing
- Yu-Sung Su's Blog
- The Numbers Guy
- Messy Matters
Visualization:
Cognitive and Behavioral Science:
- Decision Science News
- British Psychological Society Research Digest
- Seth Roberts [experimental psychology]
- Criteria's Employee Testing Blog
Social and Political Science:
- Monthly Labor Review Precis
- Marginal Revolution [economics]
- Language Log
- Social Science Statistics
- The Baby Name Wizard
- Vox EU
Machine Learning:
Cultural:
Pages
Research supported by the National Science Foundation
National Institutes of Health
Yahoo Research
Search
Recent Comments
- Aleks Jakulin: For all those lost souls that might find this through read more
Recent Entries
- The divergence of the intellectual upper class
- Inbox zero
- Big juicy datasets
- A horror story involving the correction of a published scientific article
- A question about poker
- Seniors Skeptical on Health Care Spending
- Linf: An L-infinity Classifier
- Election 2008: What really happened
- Baseball stats: innovation, randomness, and other issues
- A cool bit of experimental research on group threat and voter mobilization
- Why don't I do more explicit modeling of spatial or temporal patterns?
- Computational Legal Studies
- The work-until-its-significant strategy
- News flash: Nascar drivers support Republicans
- Two-stage regression as an approximation to multilevel modeling
- Thinking about rationality of voting, and cooperative systems in general, in terms of the design of intelligent agents
- A Central Limit Theorem Java applet
- Clustered standard errors vs. hierarchical modeling
- OmniGraphSketcher
- What do you need to apply to a Ph.D. program in statistics?
Categories
- Administrative (15)
- Art (36)
- Bayesian Statistics (239)
- Causal Inference (93)
- Decision Theory (149)
- Economics (199)
- Literature (135)
- Miscellaneous Science (120)
- Miscellaneous Statistics (423)
- Multilevel Modeling (171)
- Political Science (636)
- Public Health (133)
- Sociology (247)
- Sports (27)
- Statistical computing (121)
- Statistical graphics (152)
- Teaching (149)

For all those lost souls that might find this through Google, this is a diary of what we did:
0. The problem arose in part because Berkeley DB was used by the MovableType, and the files got corrupted because of a system outage on a weekend. The corruption was not reported by the system, but the comments didn't show in the author/administrative view of the blog.
1. Under a new account we created a new MovableType installation for performing experiments. We tried exporting the old site and importing it in a new one, but not all the relevant information (formatting, lists of authors, access permissions, etc) was transferred.
2. We used db_verify to identify corrupted database files. They were "cured" by db_dump followed by db_load (remove the original file before you recreate it with db_load).
3. While old comments could be seen, new ones couldn't. The corruption was deeper. We decided to attempt migrating the corrupted database to mySQL, hoping that the migration will read most of the good stuff, and the new database will be consistent.
4. The migration script (db2sql) crashed with the insertion error. We manually edited the source code in the DBI.pm file as not to throw an exception when the insertion is not successful, but to simply continue.
5. It seems to work, and comments that were posted in the past days are now visible. Success!
Conclusion: MovableType is too cheap.